Global Continental Changes: the Context of Palaeohydrology
Geological Society Special Publications Series Editor A. J. FLEET
GEOLOGICAL SOCIETY SPECIAL PUBLICATION NO. 115
Global Continental Changes: the Context of Palaeohydrology EDITED BY
J. B R A N S O N University of Southampton, UK A. G. B R O W N University of Exeter, UK and K. J. G R E G O R Y Goldsmiths' College, University of London, UK
1996 Published by The Geological Society London
THE GEOLOGICAL SOCIETY The Society was founded in 1807 as The Geological Society of London and is the oldest geological society in the world. It received its Royal Charter in 1825 for the purpose of 'investigating the mineral structure of the Earth'. The Society is Britain's national society for geology with a membership of around 8000. It has countrywide coverage and approximately 1000 members reside overseas. The Society is responsible for all aspects of the geological sciences including professional matters. The Society has its own publishing house, which produces the Society's international journals, books and maps, and which acts as the European distributor for publications of the American Association of Petroleum Geologists, SEPM and the Geological Society of America. Fellowship is open to those holding a recognized honours degree in geology or cognate subject and who have at least two years' relevant postgraduate experience, or who have not less than six years' relevant experience in geology or a cognate subject. A Fellow who has not less than five years' relevant postgraduate experience in the practice of geology may apply for validation and, subject to approval, may be able to use the designatory letters C Geol (Chartered Geologist). Further information about the Society is available from the Membership Manager, The Geological Society, Burlington House, Piccadilly, London WIV 0JU, UK. The Society is a Registered Charity, No. 210161. Published by The Geological Society from: The Geological Society Publishing House Unit 7, Brassmill Enterprise Centre Brassmill Lane Bath BAI 3JN UK (Orders: Tel. 01225 445046 Fax 01225 442836)
Distributors
USA AAPG Bookstore PO Box 979 Tulsa OK 74101-0979 USA (Orders: Tel. (918) 584-2555 Fax (918) 560-2652)
First published 1996
Australia The publishers make no representation, express or implied, with regard to the accuracy of the information contained in this book and cannot accept any legal responsibility for any errors or omissions that may be made. 9 The Geological Society 1996. All rights reserved. No reproduction, copy or transmission of this publication may be made without written permission. No paragraph of this publication may be reproduced, copied or transmitted save with the provisions of the Copyright Licensing Agency, 90 Tottenham Court Road, London W l P 9HE. Users registered wi,th the Copyright Clearance Center, 27 Congress Street, Salem, MA 01970, USA: the itemfee code for ~his publication is 0305-8719/96/$7.00.
Australian Mineral Foundation 63 Conyngham Street. Glenside South Australia 5065~ Australia (Orders: Tel. (08) 379-0AAA. Fax (08) 379-4634)
India Affiliated East-West Press PVT Ltd G- 1/ 16 Ansari Road New Delhi 110 002 India (Orders: Tel. (11) 327-9113 Fax (11) 326-0538)
Japan British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library. ISBN 1-897799-69-1
Typeset by Aarontype Ltd, Unit 47, Easton Business Centre, Felix Road, Bristol BS5 0HE, UK Printed by The Alden Press, Osney Mead, Oxford, UK.
Kanda Book Trading Co. Tanikawa Building 3-2 Kanda Surugadai Chiyoda-Ku Tokyo 101 Japan (Orders: Tel. (03) 3255-3497 Fax (03) 3255-3495)
Contents Preface
vii
THE CONTEXT OF PALAEOHYDROLOGY GREGORY, K. J. Introduction
1
STARKEL,L. Palaeohydrological reconstruction: advantages and disadvantages
9
ARNELL, N. W. Palaeohydrology and future climate change
19
ADAMS, J. M. & FAURE, H. Changes in moisture balance between glacial and interglacial conditions: influence on carbon cycle processes
27
WALLING,D. E. Erosion and sediment yield in a changing environment
43
BROWN, A. G. Human dimensions of palaeohydrological change
57
BAKER, V. R. Discovering Earth's future in its past: palaeohydrology and global environmental change
73
APPROACHES TO PALAEOHYDROLOGICAL ANALYSIS
Mediterranean, tropical and monsoon regions BENITO, G., MACHADO, M. J. & PEREZ-GONZALEZ, A. Climate change and flood sensitivity in Spain
85
FULLER, I. C., MACKLIN, M. G., PASSMORE, D. G., BREWER, P. A., LEWIN, J. &
WINTLE, A. G. Geochronologies and environmental records of Quaternary fluvial sequences in the Guadalope basin, northeast Spain, based on luminescence dating
99
ENZEL, Y., ELY, L. L., HOUSE, P. K. & BAKER, V. R. Magnitude and frequency of Holocene palaeofloods in the southwestern United States: a review and discussion of implications
121
THOMAS, M. F. & THORP, M. B. The response of geomorphic systems to climatic and hydrological change during the Late Glacial and early Holocene in the humid and subhumid tropics
139
KALE, V. S., ELY, L. L., ENZEL, Y. & BAKER, V. R. Palaeo and historical flood hydrology, Indian Peninsula
155
Cold regions
CARLING, P. A. A preliminary palaeohydraulic model applied to Late Quaternary gravel dunes: Altai Mountains, Siberia
165
YAMSKIKH, A. F. Late Quaternary intra-continental river palaeohydrology and polycyclic terrace formation: the example of south Siberian river valleys
181
Temperate regions HOOKE, J. M. River responses to decadal-scale changes in discharge regime: the Gila River, SE Arizona
191
vi
CONTENTS
KALICKI, T. Climatic or anthropogenic alluviation in Central European valleys during the Holocene?
205
RUMSBY, B. T. & Macklin, M. G. River response to the last neoglacial (the 'Little Ice Age') in northern, western and central Europe
217
A FUTURE FOR PALAEOHYDROLOGY
BRANSON, J., GREGORY, K. J. & CLARK, M. J. Issues in scientific co-operation on information sharing: the case of palaeohydrology
235
PILCHER, J. R. The Past Global Changes (PAGES) Project
251
BROWN, A. G. Palaeohydrology: prospects and future advances
257
Index
266
Preface This volume arises from two international meetings held at Chilworth Manor, Southampton, 9-12 September 1994 and at the Geological Society of London, 13 September 1994. The first was organized by the INQUA Commission on Global Continental Palaeohydrology (GLOCOPH) and the second by the Commission in conjunction with the British Geomorphological Research Group. The meetings each brought together researchers from a wide variety of disciplines to discuss research in global palaeohydrological change over the last 20 000 years. They were the first to be held as an integral part of the INQUA Commission. The papers in this volume are organized into three sections. The first provides an overview of the context in which palaeohydrology is studied and researched, the second reflects the approaches to palaeohydrological analysis in a number of contrasting geographical regions and the third discusses a possible future for palaeohydrology. We are very grateful to the many people who helped to organize and finance the meetings and to prepare the manuscripts, and we particularly acknowledge the assistance of those scientists who refereed the papers in this volume, and of Caroline Ensor at Goldsmiths' College. Palaeohydrology is now recognized as an important multidisciplinary area of research and we hope that this volume reflects the present achievements and provides the basis for more significant advances. Julia Branson Tony Brown Ken Gregory February 1996
Contents Preface
vii
THE CONTEXT OF PALAEOHYDROLOGY GREGORY, K. J. Introduction
1
STARKEL,L. Palaeohydrological reconstruction: advantages and disadvantages
9
ARNELL, N. W. Palaeohydrology and future climate change
19
ADAMS, J. M. & FAURE, H. Changes in moisture balance between glacial and interglacial conditions: influence on carbon cycle processes
27
WALLING,D. E. Erosion and sediment yield in a changing environment
43
BROWN, A. G. Human dimensions of palaeohydrological change
57
BAKER, V. R. Discovering Earth's future in its past: palaeohydrology and global environmental change
73
APPROACHES TO PALAEOHYDROLOGICAL ANALYSIS
Mediterranean, tropical and monsoon regions BENITO, G., MACHADO, M. J. & PEREZ-GONZALEZ, A. Climate change and flood sensitivity in Spain
85
FULLER, I. C., MACKLIN, M. G., PASSMORE, D. G., BREWER, P. A., LEWIN, J. &
WINTLE, A. G. Geochronologies and environmental records of Quaternary fluvial sequences in the Guadalope basin, northeast Spain, based on luminescence dating
99
ENZEL, Y., ELY, L. L., HOUSE, P. K. & BAKER, V. R. Magnitude and frequency of Holocene palaeofloods in the southwestern United States: a review and discussion of implications
121
THOMAS, M. F. & THORP, M. B. The response of geomorphic systems to climatic and hydrological change during the Late Glacial and early Holocene in the humid and subhumid tropics
139
KALE, V. S., ELY, L. L., ENZEL, Y. & BAKER, V. R. Palaeo and historical flood hydrology, Indian Peninsula
155
Cold regions
CARLING, P. A. A preliminary palaeohydraulic model applied to Late Quaternary gravel dunes: Altai Mountains, Siberia
165
YAMSKIKH, A. F. Late Quaternary intra-continental river palaeohydrology and polycyclic terrace formation: the example of south Siberian river valleys
181
Temperate regions HOOKE, J. M. River responses to decadal-scale changes in discharge regime: the Gila River, SE Arizona
191
vi
CONTENTS
KALICKI, T. Climatic or anthropogenic alluviation in Central European valleys during the Holocene?
205
RUMSBY, B. T. & Macklin, M. G. River response to the last neoglacial (the 'Little Ice Age') in northern, western and central Europe
217
A FUTURE FOR PALAEOHYDROLOGY
BRANSON, J., GREGORY, K. J. & CLARK, M. J. Issues in scientific co-operation on information sharing: the case of palaeohydrology
235
PILCHER, J. R. The Past Global Changes (PAGES) Project
251
BROWN, A. G. Palaeohydrology: prospects and future advances
257
Index
266
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 1-8
Introduction K. J. G R E G O R Y Goldsmiths' College, University of London, Lewisham Way, New Cross, London S14 6NW, UK Four aspects are outlined in this introduction to provide a background context for the volume. First, consideration of developments in palaeohydrology provides an opportunity to summarize the way in which the discipline has evolved. This is succeeded by a synopsis of seven prevailing themes which have been evident in the development of palaeohydrology and which have produced a number of outstanding questions. Finally the current approach to global continental palaeohydrology is explained.
Developments The first international meeting of GLOCOPH (the INQUA Commission on Global Continental Palaeohydrology) held in Southampton in September 1994 occurred 50 years after the first explicit definition of palaeohydrology was made by Leopold & Miller in 1954. Over that 50 years, interest in palaeohydrology was at first somewhat slow to develop, but Schumm in 1965 produced a major paper on Quaternary palaeohydrology in which he suggested that palaeohydrology offered an innovative approach which was capable of further exploration. The fundamental proposal of Schumm's paper (1965) was that global relationships between run-off and precipitation and also between run-off and sediment yield could be employed to indicate, for different temperature conditions, how changes might occur under different climatic conditions. This was a major breakthrough because it gave a mechanism for indicating the types of change that were possible prior to the availability of estimates of hydrological change obtained from sophisticated computer models. Further developments that occurred in the 1960s and 1970s were eventually summarized in the book by S. A. Schumm (1977) entitled The Fluvial System. That book, in addition to summarizing an innovative approach to the Quaternary, particularly through palaeohydrology, also explained an approach to river metamorphosis which previously had been developed in a number of papers (e.g. Schumm 1969) founded upon the relationship first enumerated by the hydraulic engineer, E. W. Lane (1955). A basis of analysis in quantitative stream morphology was proposed by Lane (1955) using the approximations Qsd ~ QwS where Qs is the quantity of sediment, d the particle size or size of sediment, Qw the water discharge, and S the slope of the stream. The approximation was used by Lane to demonstrate six classes of change which were used to interpret the effects of engineering works upon river channels. Gregory (1983) indicated how the foundation provided by Lane (1955) was developed in a series of important papers by S. A. Schumm culminating in his book in 1977 (Schumm 1977).
2
K.J. GREGORY
Apart from these antecedents, a major development in palaeohydrology occurred through the IGCP (International Geological Correlation Programme) Project 158 which was organized from 1978 to 1989. That project followed a programme of research organized as two sub-projects. Sub-project A, concerned with fluvial systems, was headed by Leszek Starkel and Sub-project B on bogs and mires, was led by Bjorn Berglund. The research focused upon areas in the temperate zone and was very effective in three major ways. First, it co-ordinated the results from individual palaeohydrology research investigations that were already taking place in a number of separate countries. Secondly, through the organization of regular annual meetings, often with field excursions, it was possible to effectively compare results, techniques and data to facilitate the better understanding of palaeohydrology of the temperate zone. The third successful outcome was that the project produced publications which not only advanced the techniques appropriate for palaeohydrology but also provided reports for individual countries (e.g. Starkel 1990; Gregory et al. 1987) and gave correlated results from the overall project. The final volume, entitled Temperate Palaeohydrology, was published in 1991 (Starkel et al. 1991) and brought together results from 14 countries in the temperate zone.
Prevailing themes Palaeohydrology may be defined as the science of the waters of the earth, their composition, distribution and movement on ancient landscapes from the beginning of the first rainfall to the beginning of continuous hydrological records (Schumm 1977; Gregory 1983). As research has progressed, it has become apparent that a number of separate themes has been pursued in palaeohydrology and that particular schools of approaches have been fostered. Perhaps as many as seven approaches might be identified from the research from 1965 to 1991, although inevitably there is overlap between each of the themes. First, there have been studies of cut and fill sequences and of terrace and valley floor development. Analyses of valley in-fills have allowed the interpretation of the palaeohydrologic conditions under which different stages of valley floor development occurred. This was the raison d~tre for the study by Leopold & Miller (1954) that led to the first formal definition of palaeohydrology. More recently other models of flood-plain and valley floor development have been developed such as the alternation of episodes of vertical accretion and catastrophic stripping proposed by Nanson (1986). Whereas that general approach was emphasised initially in semi-arid areas such as the south west of the United States, a second approach was essentially based upon the water balance (Schumm 1965) and was capable of application to a range of environments and particularly to temperate areas that had experienced periglacial conditions during the Quaternary. This approach embraced modelling techniques as linked to climatic change (e.g. Lockwood 1983) although the inclusion of palaeohydrology as an integral part of global change programmes has not been readily accomplished (Gregory 1995). A third approach was initiated by Dury (1964a, b, 1965) when he analysed former large meanders based upon valley meanders, concluded that extensive stream shrinkage had occurred, and deduced that one of the major reasons for such
INTRODUCTION
3
shrinkage was change of climate. Studies of such underfit streams were conducted by Dury and later developed by other researchers in the context of particular areas (e.g. Rotnicki 1983, 1991; Maizels & Aitken 1991). These investigations led to the conclusions that, not only had river discharges been significantly greater in the past, but also morphometric analysis of contemporary meanders and comparison with valley meanders could facilitate estimation of the palaeodischarges that had occurred. Allied to this approach were studies based upon palaeohydraulic methods which utilized analyses of sediment characteristics to furnish information about palaeodischarges and palaeohydrologic conditions (Gregory & Maizels 1991; Williams 1983). The diversity of approaches and the equations available were stressed by Williams (1984, 1988). An approach to river metamorphosis can be identified as a fourth theme and was developed in a series of research investigations (e.g. Schumm 1969, 1977) which served to demonstrate the extent to which, in the Holocene, there had been a significant number of major changes of river channel patterns instigated by climatic changes and by human activity, with the influence of the two causes often interwoven in a complex way. This complexity has been demonstrated in studies of the upper Mississippi (Knox 1984), in Europe in general and in Poland in particular (Starkel 1991). Results obtained from the investigations of bogs and mires, particularly using palaeoecological techniques, provide a fifth strand of palaeohydrologic research and not only was this a theme in IGCP Project 158, Sub-project B but it has been the basis for a considerable number of research investigations (Berglund 1979) which have led to conclusions about the water balance under previous climatic conditions which in turn have been the basis for estimating the characteristics of the palaeohydrology (Berglund 1977, 1983; Lang & Schluchter 1988). In some areas, lake fluctuations have provided a significant indicator of hydrological changes and so lake databases (e.g. Harrison 1988; Street-Perrott et al. 1985) have been employed in a sixth approach to give information about the characteristics of palaeohydrological environments. These six approaches have been used to varying degrees in studies of particular basins in the temperate zone to enable reconstruction of palaeohydrological changes during the last 15 000 years as part of IGCP Project 158. Thus has arisen a seventh approach which has endeavoured to integrate individual approaches and which includes impressive studies undertaken in the Vistula basin (Starkel 1995) and in other basins investigated as part of the project and described in Temperate Palaeohydrology (Starkel et al. 1991). The seven types of approach were evident by the early 1980s and as research has progressed it has been aided by the application of new techniques including palaeomagnetism (Oldfield 1983), the use of trace minerals and mining debris (Lewin & Macklin 1986), and also by the investigation of palaeostage indicators (PSI) which have been employed in a variety of areas to give information on former catastrophic floods. This work was initially set in the context of flood hydrology (Baker et al. 1988) and a series of impressive research investigations has been achieved utilizing evidence from slack water deposits and from their situation and sequence (Baker 1987). In addition, other developments have occurred especially in relation to groundwater changes, in particular the effects of glaciers on groundwater (Boulton & Spring 1986; Boulton & Dobbie 1993; Boulton et al. 1993) and changes in arid regions (Issar et al. 1984; Issar 1985).
4
K . J . GREGORY
Outstanding questions The general achievement of research over the last 50 years has enabled the improved understanding of palaeohydrological environments in the light of knowledge of contemporary hydrological processes. Thus, as the understanding of hydrology and of fluvial processes has advanced since the 1960s, particularly with the development of the variable source area concept and run-off modelling (e.g. Gregory & Walling 1973), the investigation of palaeoenvironments has now benefited enormously from an enhanced understanding of contemporary environmental processes. This awareness of global environmental process is reflected in recent summary texts (e.g. Mannion 1991; Roberts 1994) although the potential contribution of palaeohydrological research in the study of global environments is not always as clearly acknowledged as it might be (Gregory 1995). Although global change research programmes have developed significantly over the decade since 1980, and many now acknowledge the contribution that palaeohydrology can make, nevertheless palaeohydrology has yet to attract the level of funding that has been associated with general circulation models, with atmospheric modelling or with analyses of major oceanographic systems. This is perhaps slightly paradoxical because the impact of global change is often reflected through the river system so that the knowledge of the magnitude of palaeohydrological river system changes could have significant implications for understanding river flows in relation to human activity and hazards. However, not all areas of the world have been studied in equal detail using palaeohydrological methods of any kind. Research investigations included in a database (Branson et al. 1995), which when analysed according to world distribution (Fig. 1), show that there has been a preponderance of research in European temperate areas and in some parts of North America. In just the same way that Graf (1984) related the location of geomorphological research investigations to the
,.la~
.
o
Fig. 1. Location of areas which are the subject of published research (English language) on palaeohydrology to 1994. The distribution was produced from a database compiled by Julia Branson (see pp. 235) funded by a Leverhulme Research Grant to K. J. Gregory.
INTRODUCTION
5
different types of environment in North America, so it is possible to see how the themes that have prevailed in the development of palaeohydrology research have been associated with particular areas of the world. Despite the concentration of research in western and central Europe and to a lesser degree in the United States (Fig. 1), it is encouraging to see how some investigations have been undertaken in other continents but there is an outstanding need for further studies to increase the understanding of palaeohydrological change; the chapters in this volume and research work as part of GLOCOPH should fill some of the gaps. A further outstanding requirement is to place the results of palaeohydrological investigations in a functional drainage basin framework. Contemporary hydrological models have to be developed for drainage basins, and to further advance modelling in palaeohydrology it is desirable to analyse the research results within the drainage basin framework. Although this was originally advocated by Schumm (1977) it remains to be fully developed; research of this kind is a theme which needs to be kept in mind with the further development of palaeohydrological research. Allied to progress in this direction will be the application of significant recent results in interpreting the geomorphic effectiveness of different kinds of floods (e.g. Costa & O'Connor 1995) and the spatial patterns of erosion and deposition (e.g. Miller 1995) to palaeohydrological sequences as anticipated by Thornes & Gregory (1991) and recognized by Petts (1995) and Gardiner (1995). This approach may enable us to add further to present understanding of changes of river channels during the Holocene (Starkel 1995) and provide what Baker (1991) called a bright future for old flows.
An approach to global continental palaeohydrology Because of the variations in the intensity of palaeohydrological research (Fig. 1) and also the need to correlate results from one area to another, it was desirable to establish a collaborative research programme that synthesized results, from a variety of world areas, obtained by utilizing a range of techniques. At INQUA, in Beijing in 1991, a new commission was established concerned with global continental palaeohydrology (GLOCOPH) with Leszek Starkel as President and V. R. Baker and K. J. Gregory as Vice Presidents. It was intended that this new commission would build upon the work already achieved by the IGCP project, facilitate the investigation of palaeohydrology on a global basis and place the results in a context related to global water balance changes (e.g. Waylen 1995). Whereas the IGCP project had raised the profile of palaeohydrology and increased the awareness of the contribution that this interdisciplinary field could make, it was also timely to relate studies of contemporary processes to past environments and to use the increased understanding gained for the elucidation of future environments. The Global Continental Palaeohydrology Commission of INQUA was established in 1991 potentially using the involvement of some 90 researchers in more than 30 countries. The first international meeting of the Commission held in Southampton in 1994 involved the presentation of a series of important papers and this was followed by a day of lectures at the Geological Society of London when several invited contributions demonstrated the context for palaeohydrology. It is from those papers presented in London on 13 September and in Southampton between 10 and 12 September 1994 that this volume has been compiled. As palaeohydrology can seek to improve our understanding of past environments using a knowledge of
K.J. GREGORY
6
contemporary processes, the research undertaken in this way is able to complement knowledge gained from the period of instrumented records and it may in turn provide insight into environments that could be analogues for future situations.
References BAKER, V. R. 1987. Palaeoflood hydrology and extraordinary food events. Journal of Hydrology, 96, 79-99. - - 1 9 9 1 . A bright future for old flows. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, 497-520. , KOCHEL, R. C. & PATTON, P. A. (eds) 1988. Flood Geomorphology. Wiley, New York. BERGLUND, B. E. 1977. Handbook Of Holocene Palaeoecology and Palaeohydrology. Wiley, Chichester. - - 1 9 7 9 . Presentation of the IGCP Project 158b Palaeohydrological Changes in the Temperate Zone In The Last 15,000 Years- Lake And Mire Environments. Acta Universitatis Ouluensis, Series A, 82, 39-48. - - 1 9 8 3 . Palaeohydrological studies in Lakes and Mires- A Palaeoecological Research Strategy. In: GREGORY, K. J. (ed.) Background to Palaeohydrology. Wiley, Chichester, 237-254. BRANSON, J., CLARK, M. J. & GREGORY, K. J. 1995. A Database For Global Continental Palaeohydrology: Technology or Scientific Creativity? In: GREGORY, K. J., STARKEL, L. & BAKER, V. R. (eds) Global Continental Palaeohydrology. Wiley, Chichester, 303-321. BOULTON, G. S. & DOBBIE, K. E. 1993. Consolidation of sediments by glaciers; Relations between sediment geotechnics, soft-bed glacier dynamics and sub-glacial ground-water flow. Journal of Glaciology, 39, 26-44. - - & SPRING, U. 1986. Isotopic Fractionation at the base of Polar and Sub-polar Glaciers. Journal of Glaciology, 32, 475-485. , SLOT, T., BLESSING, K., GLASBERGEN, P., LEIJNSE, T. & VANGIJSSEL, K. 1993. Deep Circulation of groundwater in overpressured subglacial aquifers and its geological consequences. Quaternary Science Reviews, 12, 739-745. COSTA, J. E. & O'CONNOR, J. E. 1995. Geomorphically effective floods. In: COSTA, J. E., MILLER, A. J., POTTER, K. W. & WlLCOCK, P. R. (eds) Natural and Anthropogenic Influences in Fluvial Geomorphology. American Geophysical Union Geophysical Monographs, 89, 45-56. DURY, G. H. 1964a. Principles Of Underfit Streams. US Geological Survey Professional Paper 452A. - - 1 9 6 4 b . Subsurface Explorations and Chronology of Underfit Streams. US Geological Survey Professional Paper 452B. - - 1 9 6 5 . Theoretical Implications of Underfit Streams. US Geological Survey Professional Paper 452C. GARDINER, V. 1995. Channel Networks: progress in the study of spatial and temporal variations of drainage density. In: GURNELL, A. & PETTS, G. (eds) Changing River Channels. Wiley, Chichester, 65-85. GRAF, W. L. 1984. The geography of American field geomorphology. Professional Geographer, 36, 78-82. GREGORY, K. J. (ed.) 1983. Background to Palaeohydrology; A Perspective. Wiley, Chichester. - - 1 9 9 5 . Human activity and palaeohydrology. In: GREGORY, K. J., STARKEL, L. & BAKER, V. R. (eds) Global Continental Palaeohydrology. Wiley, Chichester, 151-172. & MAIZELS, J. K. 1991. Morphology and sediments: typological characteristics of fluvial forms and deposits. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, Chichester, 31-59. - - & WALLING, D. E. 1973. Drainage Basin Form and Process. Edward Arnold, London. , LEWIN, J. & THORNES, J. B. 1987. Palaeohydrology In Practise. Wiley, Chichester. HARRISON, S. P. 1988. Reconstructing climatefrom lake level changes. Lundqua Thesis 21, Lund. -
-
INTRODUCTION
7
ISSAR, A. 1985. Fossil water under the Sinai-Negev Peninsula. Scientific American, 153, 104-110. --, KARNIELI, A., BRUINS, H. J. & CASTRO, M. 1984. The Quaternary geology and hydrology of SedeZine, Negev, Israel. Israel Journal of Earth Sciences, 33, 34-42. KNOX, J. C. 1984. Responses of river systems to Holocene climates. In: WRIGHT, H. E. (ed.) Late Quaternary Environments of the United States. Volume 2 The Holocene. Longman, 26-41. LANE, E. W. 1955. The importance of fluvial morphology in hydrologic engineering. Proceedings American Society of Civil Engineers, 81, Paper 745, 1-17. LANG, G. & SCHLUCHTER,C. 1988. Lake, Mire and River Environments during the last 15,000 Years. A. A. Balkema, Rotterdam. LEOPOLD, L. B. & MILLER, J. P. 1954. Postglacial chronology for alluvial valleys in Wyoming. United States Geological Survey Water Supply Papers, 1261, 61-85. LEWlN, J. & MACKLIN, M. G. 1986. Metal mining and flood plain sedimentation in Britain. In: GARDINER, V. (ed.) International Geomorphology, 1986 Part 1. Wiley, 1009-1027. LOCKWOOD, J. G. 1983. Modelling climatic change. In: GREGORY, K. J. (ed) Background to Palaeohydrology. Wiley, Chichester, 25-50. MAIZELS, J. K. & AITKEN, J. 1991. Palaeohydrological change during deglaciation in Upland Britain: A case study from North-East Scotland. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, Chichester, 105-145. MANNION, A. M. 1991. Global Environmental Change. Longman, Harlow. MILLER, A. J. 1995. Valley morphology and boundary conditions influencing spatial patterns of flood flow. In: COSTA, J. E., MILLER, A. J., POTTER, K. W. & WILCOCK, P. R. (eds) Natural And Anthropogenic Influences in Fluvial Geomorphology. American Geophysical Union, Geophysical Monographs, 89, 57-81. NANSON, G. C. 1986. Episodes of vertical accretion and catastrophic stripping: A model of disequilibrium flood plain development. Geological Society of America Bulletin, 97, 1467-1475. OLDFIELD, F. 1983. The role of magnetic studies in palaeohydrology. In" GREGORY, K. J. (ed.) Background To Palaeohydrology: A Perspective. Wiley, Chichester, 141-165. PETTS, G. E. 1995. Changing river channels: the geographical tradition. In: GURNELL, A. & PETTS, G. E. (eds) Changing River Channels. Wiley, Chichester, 1-23. ROBERTS, N. (ed.) 1994. The Changing Global Environment. Blackwell, Oxford. ROTNICKI, A. 1983 Modelling past discharges of meandering rivers. In: GREGORY, K. J. (ed.) Background To Palaeohydrology: A Perspective. Wiley, Chichester, 321-354. - - 1 9 9 1 . Retrodiction of palaeodischarges of meandering and sinuous alluvial rivers and its palaeohydroclimatic implications. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology. Wiley, Chichester. SCHUMM, A. 1965. Quaternary palaeohydrology. In: WRIGHT, n. E. & FREY, D. G. (eds) The Quaternary Of The United States. Princeton University Press, Princeton, 783-794. - - 1 9 6 9 . River metamorphosis. Proceedings American Society of Civil Engineers. Journal of Hydraulics Division, 6352, HY1,255-273. - - 1 9 7 7 . The Fluvial System. Wiley, Chichester. STARKEL, L. 1990. Fluvial environment as an expression of geological changes. Zeitschriftffir Geomorphologie, Supplementband, 79, 133-152. - - 1 9 9 1 . The Vistula river valley: a case study for central Europe. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, Chichester, 171-188. - - 1 9 9 5 . Changes of river channels in Europe during the Holocene. In: GURNELL, A. & PETTS, G. (eds) Changing River Channels. Wiley, Chichester, 27-42. --, GREGORY, K. J. & THORNES, J. B. (eds) 1991. Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, Chichester. STREET-PERROTT, A., ROBERTS, N. & METCALFE, S. 1985. Geomorphic implications of late Quaternary hydrological and climatic changes in the northern hemisphere Tropics. In: DOUGLAS, I. & SPENCER, T. (eds) Environmental Change and Tropical Geomorphology. George Allen & Unwin, London, 165-183.
8
K.J. GREGORY
THORNES, J. B. & GREGORY, K. J. 1991. Unfinished business: a continuing agenda. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15,000 Years. Wiley, Chichester, 521-536. WAYLEN, P. 1995. Global hydrology in relation to palaeohydrological change. In: GREGORY, K. J., STARKEL, L. & BAKER, V. R. (eds) Global Continental Palaeohydrology. Wiley, Chichester, 61-86. WILLIAMS,G. P. 1983. Palaeohydrological methods and some examples from Swedish fluvial environments, 1. cobble and boulder deposits. Geografiska Annaler, 65A, 227-243. - - 1 9 8 4 . Palaeohydrologic equations for rivers: equations and methods. In: COSTA, J. E. & FLEISCHER, P. J. (eds) Developments and Applications Of Geomorphology. SpringerVerlag, Berlin, 343-367. - - 1 9 8 8 . Palaeofluvial estimates from dimensions of former channels and meanders. In: BAKER, V. R., KOCHEL, R. C. & PATTON, P. C. (eds) Flood Geomorphology. Wiley, New York, 321-334.
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 9-17
Palaeohydrological reconstruction: advantages and disadvantages L. S T A R K E L
Institute of Geography, Polish Academy of Sciences, 31-018 Krakow, ul sw. Jana 22, Poland Abstract: The study of continental palaeohydrology is a young branch of science which has drawn on knowledge of the mechanisms, and observations, of presentday processes to reconstruct past hydrological conditions. Patterns of annual precipitation and evaporation, for example, can be determined from the distribution of plant communities or by dendrochronological techniques, lake-level variations can provide an indication of water storage fluctuations, reconstructions of runoff, particularly bankfull discharge and mean annual discharge can be based on palaeochannel form and sedimentology, and extreme floods can be studied using slack-water deposits. All of these techniques rely on the acceptance of assumptions which may not always be valid. It is therefore important that several different methods are used to provide a collaborative approach both to validating methods and in determining the driving force of hydrological changes.
In many palaeogeographic and palaeoclimatic reconstructions we touch upon the components of the hydrological cycle, mainly precipitation and an index of aridity. The water balance equation is the basis of these reconstructions, the elements of which may-be retrodicted on the basis of geological evidence (stratigraphy, forms, fossil flora and fauna) as well as with models (Fig. 1).
changes of ice volume, snow storage sea, lake level cnan.ges, rate of peat arowth changes of g~ound water level
tree rings landslides .,~r
,,u~
. , 4 . . , ~
~
~'~ ""~
' " "
'~B' ' "
P=E
I
' ~ ~
qmr ~
"ram ~
~
~ ~
§
vegetation type so!! ty.pe .. sail oeposlts
~
~
+
"
S
dated .clround waters soiis(gTeying, leaching)
paleochannels . aluvial facies . , I+ cataclysmic flooos) Fig. 1. Equation of the water balance. Main sources of information on the paleohydrological parameters.
10
L.
S T A R K E L
At the outset the validity of the information of the water balance elements used to reconstruct past water circulation should be questioned (Starkel 1993). The reconstructions make the picture of changes more rational. The amount of water stored in the oceans and lakes are the most appropriate factors for quantification. However, it should be recognized that calculations of the present water balance differ significantly (e.g. Keller 1962; Lvovitch 1974; Shiklomanov 1980), and errors in the calculation of total water resources, runoff and atmospheric precipitation may be very large indeed.
Sources of information about hydrological parameters If the water requirements of plant species are known, then on the basis of the uniformitarian approach it is possible to reconstruct the spatial and temporal patterns of precipitation and evapotranspiration (Webb & Bryson 1972; Grichuk et al. 1984; Klimanov 1990), and indirectly runoff (Georgiadi 1992) (Figs 2 & 3).
: .~..:-:-.-. ._ . - . - _ - - . , -- - - 9-
.
-
-
.-.
-...
-, - . - ; . - .
.
.
.
.
.
t
0
50
F:::::::-I
IIXI
200 mm
I
11
no .Jmoll
Fig. 2. Annual mean runoff during the Last Interglacial climatic optimum (deviations from present day values) (after Georgiadi 1992).
PALAEOHYDROLOGICAL RECONSTRUCTION
11
A problem with these reconstructions, however, is that the zonal concept is often over-simplified; ecotonal zones have shifted and varied in time, and the statistical approximations may be insufficient to accurately describe the ecotonal response to water deficit or permafrost conditions (Webb 1983). Under these environmental conditions the ecotonal zones may be extremely wide and complicated (cf. Frenzel et al. 1992), and the locations where organic material has deposited are rare and sitespecific and therefore not representative of large areas. Thus, in the case of permafrost regions, the calculations of annual precipitation totals can be overestimated. Variations in lake levels have provided the best indicators of long-term trends in mean annual rainfall and water storage. Changes in lake level are reflected in coastal forms, facies differentiation and faunal spectra of diatoms and cladocera (e.g. StreetPerrott & Harrison 1985; Digerfeldt 1986). The most widely accepted reconstructions have been obtained for closed depressions of tectonic or similar origin (e.g. the Caspian Sea; Klige & Myagkov 1992).
iii.o
...-" -
.0 .....
ooo0
~
.
~
.o
.A
q
./7
_
.,"a,,,
~
,c~
-----_--__-_-._---_-__-~
..00.. o..0.o
A
q
-
.
.
o
r3:t-b
,e
. . . .
-I00
-50
0
-..-..-
50
-..-_-
_
.
.
I00 mm
.
.
no nmolf
Fig. 3. Annual mean runoff during the Holocene climatic optimum (deviations from present day values) (after Georgiadi 1992).
12
L. STARKEL
Fluctuations of lake level and water input have been reconstructed for the North Crimean Lake Saki based on the thickness and lithology of the annual laminae. These fluctuations have been correlated with the discharge of the Dnieper River for the last 2500 years (Schwetz 1978). In the peat cores breaks in peat stratigraphy indicate phases of desiccation or drought, e.g. of distinct lowerings of the groundwater level (Casparie 1972; Aaby 1976). Many of these droughts occurred only locally, however, and do not provide a regional picture of hydrological change. In deglaciated areas or thermokarstic depressions, the water level may be controlled by long-term trends in lake drainage or overtopping and thus the influence of climatic factors may only be indirect. Calculations of the percentage of randomly investigated lakes with higher or lower water levels in particular time slices may thus lead to incorrect conclusions (Harrison et al. 1993). Reconstructions of runoff, particularly of bankfull discharge and mean annual discharge, are based on parameters derived from palaeochannel shape and sediments. Unfortunately, however, identification of the majority of these parameters is problematic (Gregory & Maizels 1991; Soja 1994). There are many equations that can be used to calculate bankfull discharges and mean annual discharges from measurements of meandering channels (Dury 1977; Rotnicki 1991). Most of these equations are based on the assumptions presented below. In the case of palaeomeanders the existence of a single thread channel is assumed to be the normal equilibrium planform. It is difficult to determine the bankfull discharge or the channel depth and the base of the coarse (armoured) horizon from stratigraphical evidence: there may be several such layers and the upper one may have been formed after the channel was cut off. Mature, tortuous free meanders were probably created during long, relatively stable periods (perhaps lasting centuries or millennia), but conversely, the reconstructed discharge is related to the moment just preceding the beginning of filling dated by radiocarbon techniques. In the case of braided palaeochannels the errors may be even greater that those involving single-thread channels, and may reach several hundred percent (Maizels 1983), as it is difficult to determine which branches of the braided system were active at any one point. Moreover, the grain size composition of the channel facies may provide ambiguous information on the flood discharges (Church 1978). Extreme floods are studied by the analysis of coarse channel debris and slackwater deposits (e.g. Baker 1987). Using these techniques the occurrence of cataclysmic floods from a variety of origins has been identified on different continents (Baker et al. 1988; Rudoy & Baker 1993). When reconstructing the frequency of extreme events on this basis, however, the preliminary assumptions should be recalled. In one sequence of slackwater deposits several floods may be identified but each subsequent event must exceed the previous one in order to be registered in the stratigraphic sequence. Even in the case of the famous Missoula flood it is not known whether it was a single or several events, close or distant in time (Baker & Bunker 1985). Additionally, it may have been assumed that the channel floor was cut into the bedrock, an0 was therefore stable. Great floods may have deposited several metres of debris on the rocky bottom which were removed by a later flood event (cf. the Tista river in the Sikkim Himalayas; Froelich & Starkel 1987). New methods of dating historical alluvia by the use of heavy metals and radioactive elements (Zn, Pb, Cs etc.) helps to identify single floods and evaluate the more precise time intervals between them (Knox 1983; Macklin & Dowsett 1989).
PALAEOHYDROLOGICAL RECONSTRUCTION
13
In the temperate zone a clustering of contemporary black oaks is considered as an indicator of frequent flooding (Becker 1982). The use of such indicators requires caution since these trunks, as well as the sediments accompanying them, may be redeposited depending upon the local conditions (Kalicki & Krapiec 1995). Extreme rainfalls (heavy downpours, continuous rains and rainy seasons; cf. Starkel 1976) are often accompanied by mass movements (e.g. debris flows and earthflows, shallow- and deep-seated landslides). Their distribution in time may be essentially random and therefore correlation with sequences of other events is needed (Starkel 1985, in press). Most mass movements are dated using organic deposits which fill the depressions over the sliding masses; such dating methods may underestimate the age of the landslide, however, due to the lag in peat formation or lacustrine deposition. The most reliable datings originate from sediments overridden by sliding masses or from reservoirs dammed by landslides (Alexandrowicz in press). In the cold regions changes in the radiation/evaporation balance and snowfall are reflected by fluctuations of glaciers. A careful examination of the advance and retreat of Alpine glaciers indicates that parallel tendencies can be observed. An individual glacier, however, reacts to variations in temperature and precipitation with different intensities and response times (Patzelt 1985). Therefore fluctuations of the ice fronts can only adequately inform about the centennial changes. Datings of fossil reservoirs, particularly in sedimentary rocks in the arid zone can provide information regarding periods of high groundwater storage (Geyh 1972; Sonntag et al. 1981). Again, interpretation must be cautious, since the reservoirs may be of a relict character or subject to manifold or continuous mixing.
Reconstruction of the water budget The methods described above for the reconstruction of components of the water cycle or water budget serve as tests of retrodictions of the whole water balance. Closed lacustrine catchments are the most suitable environment to undertake this type of analysis. It is possible to compare geological records with the retrodiction based on modelling (Kutzbach 1980, 1992; Swain et al. 1983; Klige & Myagkov 1992). More difficult and uncertain are the tests of balance calculations based only on the water requirements of vegetation (Georgiadi 1992) or on the runoff characteristics reconstructed from the channel parameters (Rotnicki 1991). Different seasonal distributions of rainfall and type of flooding are connected with snow storage and existence of permafrost, and if these factors are not taken into consideration for the Late Glacial period the retrodicted precipitation and evaporation may be overestimated (Klimanov 1990; Rotnicki 1991). Thechanges reconstructed from palaeohydrological analysis may be induced by climatic change and/or by modification of the water cycle by human activity. The relationship between flood frequency and rise of groundwater level with forest clearance, for example, are generally well known. The evidence gathered from valley floors in Poland (Starkel 1991b; Kalicki 1991) and in Britain (Needham & Macklin 1992), for example, indicate that the youngest observed phases of aggradation or higher flood frequency were caused by the coincidence of the climatic phases and human intervention (e.g. the late Roman phase, termination of the Medieval period).
14
L. STARKEL
In undertaking palaeohydrological reconstructions a special role is played by the interpretation of radiocarbon datings. The dated organic samples may either be assigned to the sediments in situ or to the re-deposited sediment. The histograms of frequency of the organic samples, used as the indicators of a higher water level or higher precipitation, may lead to incorrect conclusions. Especially doubtful are the reconstructions for the periods of time which correspond to rapid changes in vegetation and climate 14C method is not sufficiently precise. A deviation of 100 or 200 years in the time intervals of 6000 or 5000 BP may result in the correlation of different phases or events. In such cases the different response times of different geoecosystems to long-term trends or extreme events should be considered.
Discussion and conclusions When palaeogeographic reconstructions, including palaeohydrological ones, are made for specific time intervals it is assumed that the geoecosystems are in equilibrium with the radiation balance. This assumption, however, is erroneous at least for the beginning of the Holocene and for the Atlantic-Subboreal transition. There are also other causes of diachronity, for example, the stable ecosystems and fluvial systems of lowland areas demand much higher threshold values to exceed equilibrium conditions than more unstable mountain geoecosystems (Starkel 1991a). The critical evaluation of the evidence and methods presented above does not imply that palaeohydrological reconstructions are unachievable. On the contrary, there is a strong need for the careful comparison of hydrological changes registered in sequences of sediments, in order to explain the causes and mechanisms of change. Several recent approaches (e.g. Harrison et al. 1993) have shown that the results obtained by different methods are difficult to correlate. This implies that either the methods used are incorrect or incomparable or the derived conclusions may be wrong (due to equivocal assumptions). Two kinds of records of hydrological change in the past should be distinguished. First, continuous records of change in relatively stable environments such as laminated lacustrine sediments, cave calcareous precipitation, peat bogs, soil profiles, tree rings or ice cores. All of these register undisturbed long-term trends of changes in the water budget. Second, records which register extreme changes caused by fluctuations in precipitation and runoff such as catastrophic rainfall, flooding and drought. To retrodict a hydrological regime it is necessary to infer from both kinds of records (Klige 1990) which may simultaneously register different causes of change. In the temperate zone of Eurasia there are, for example, long-term changes in water storage which reflect changes in the air mass circulation pattern. This is observed in the N-S transect across Europe, which shows diachronous changes of lake levels and glacial advances during the Holocene (Karlen 1991; Harrison & Digerfeldt 1993). On the other hand, phases with a high frequency of extreme events have much wider coverage and are evidenced in tree ring patterns, flood deposits etc. connected with random rhythmicity of different global or hemispheric origins (e.g. phases of high volcanic activity; Nesje & Johannessen 1992). The correlation between phases of different humidity and water storage with periods of extreme hydrological events indicate the existence of regions where both kinds of rhythmicity are superimposed or in phase with each other, e.g. wet phases coincide with phases of higher flood frequency. During the Holocene such
PALAEOHYDROLOGICAL RECONSTRUCTION
15
coincidence occurred in Central Europe. In other parts of the globe these two types of rhythmicity were probably out-of-phase. In many cases it is difficult to separate the long-term changes in the water budget from single extreme events. In closed lake basins a similar effect in the form of a rise in a lake level in areas of water deficit may be caused either by a secular rising trend of precipitation and runoff (or decline of evaporation) or by rapid catastrophic rain (cf. heavy rain at Sambhar Lake in Rajasthan; Starkel 1972). The study of continental palaeohydrology is a young branch of science which has drawn on knowledge of the mechanisms and observations of present-day processes. The methods of reconstruction require further testing and continuous parameterisation, and their imperfections should be verified by cross-correlation of records obtained by alternative methods.
References AABY,B. 1976. Cyclic climatic variations in climate over the past 5500 years reflected in raised bogs. Nature, 263, 281-284.
ALEXANDROWICZ,S. W. Holocene landslides in the Polish Carpathians. In: FRENZEL,B. (ed.) ESF workshop on rapid mass movements, Mainz. Palaeoklimaforschung, in press.
BAKER, V. R. 1987. Paleoflood hydrology and extraordinary flood events. Journal of Hydrology, 96, 79-99. & BUNKER, R. C. 1985. Cataclysmic late Pleistocene flooding from Glacial Lake Missoula: A review. Quaternary Science Review, 4, 1-41. - - , KOCHEL, R. C. & PUTTON, P. C. (eds) 1988. Flood Geomorphology. J. Wiley. BECKER, B. 1982. Dendrochronologie und Palaoekologie subfossiler Baumsta'mme aus Flussablagerungen, ein Beitrag zur nacheiszeitlichen Auenentwicklung im sudlichen Mitteleuropa. Mitteilungen der Kommission fur Quart/irforschung, (3esterreich Akadamie der Wissenschaften, 5, Wien. CASPARIE,W. A. 1972. Bog Development in Southeastern Drenthe (The Netherlands). Rijks Univ. Groningen. CHURCH, M. 1978. Palaeohydrological reconstructions from a Holocene valley fill. Fluvial Sedimentology, 5, 743-772. DIGERFELDT, G. 1986. Studies on past lake-level fluctuations. In: BERGLUND,B. E. (ed.) Handbook of Holocene Palaeoecology and Palaeohydrology. J. Wiley, 127-143. DURY, G. U. 1977. Underfit streams: retrospect, perspect and prospect. In: GREGORY,K. J. (ed.) River Channel Changes. Wiley, Chichester, 281-293. FRENZEL, B., PECSI, M. & VELICHKO, A. A. 1992. Atlas ofpaleoclimates and paleoenvironments of the Northern Hemisphere, Late Pleistocene-Holocene. Hungarian Academy of Sciences, G. Fischer Verlag, Budapest-Stuttgart. FROEHLICH,W. & STARKEL,L. 1987. Normal and extreme monsoon rains--their role in the shaping of the Darjeeling Himalaya. Studia Geomorphologica Carpatho-Balcaniea, 21, 129-160. GEORGIADI,V. 1992. Paleohydrological reconstructions. In: FRENZEL,B. et al. (eds)Atlas of Paleoclimates and Paleoenvironments of the Northern Hemisphere. Hungarian Academy of Science, G. Fischer Verlag, Budapest-Stuttgart. GEYH, M. A. 1972. On the determination of the initial 14C content in ground water. In: RAFTER, T. A. & GRANT-TAYLOR, T. (eds) 8th International 14C Conference. Royal Society of New Zealand, Wellington, D58-D69. GREGORY,K. J. & MAIZELS,J. 1991. Morphology and sediments: tyopological characteristics of fluvial forms and deposits. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology. J. Wiley, 31-59. GRICHUK, V. P., GURTOVAYA,Y. Y., ZELIKSON, E. M. & BORISOVA, O. K. 1984. Methods and results of Late Pleistocene Paleoclimatic Reconstructions. In: VELICHKO, A. A. (ed.) Late Quaternary Environments of the Soviet Union. Minnesota University Press, 251-260.
16
L. STARKEL
HARRISON, S. & DIGERFELDT, G. 1993. European lakes as paleohydrological and palaeoclimatic indicators. Quaternary Science Review, 12, 233-248. - - , PRENTICE, I. C. & GUIOT, J. 1993. Climatic controls of the Holocene lake-level changes in Europe. Climatic Dynamics, 8, 189-200. KALICKI, T. 1991. The evolution of the Vistula river valley between Cracow abnd Niepolotmice in Late Vistulian and Holocene times. Geographical Studies, Special Issue, 6, Warszawa, 11-37. & KR~PIEC, M. 1995. Problems of dating alluvium using buried subfossil tree trunks: lesson from the "black oaks" of the Vistula valley, central Europe. The Holocene, in press. KARLEN, W. 1991. Glacier fluctuations in Scandinavia during the last 9000 years. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology. J. Wiley, 395-412. KELLER, R. 1962. GewCisser und Wasserhaushalt des Festlandes. Jeubner Verlag, Leipzig. KLIGE, R. K. 1990. Historical changes of the regional and global hydrological cycles. Geo-Journal, 20, 129-136. & MYAGKOV, M. S. 1992. Changes in the water regime of the Caspian Sea. Geojurnal, 27, 299-307. KLIMANOV, W. A. 1990. Kolichestvennye characteristiki klimata severnoy Ewrazji w pozdnelednikovye. [Quantitative characteristics of the Northern Eurasia in the Lateglacial] [in Russian]. Izvestiya Akademii Nauk SSSR, Seria Geografiya, 4, 116-126. KNOX, J. C. 1983. Responses of river systems to Holocene climates. In: WRIGHT, H. (ed.) Late-Quaternary Environments of the United States. University of Minnesota Press, 26-41. KUTZBACH, J. E. 1980. Estimates of past climate of paleolake Chad North Africa, based on a hydrological and energy-balance model. Quaternary Research, 14, 210-223. - - 1 9 9 2 . Modeling earth system changes of the past. In: OJIMA,D. (ed.) Modeling the Earth System. UCAR, OILS Boulder, Colorado, 376-404. LVOVITCH, M. I. 1974. Wodnyje resursy mira i ich buduscije [Water resources of the Earth and their future] [in Russian]. Mysl, Moskwa. MACKLIN, M. G. & DOWSETT, 1989. The chemical and physical speciation of trace metals in fine grained overbank flood sediments in the Tyne basin, NE England. Catena, 16, 135-151. MAIZELS, J. K. 1983. Channel changes, paleohydrology and deglaciation: evidence from some lateglacial sandur deposits, northeast Scotland. Proceedings of the Holocene Symposium, Quaternary Studies in Poland, 4, Poznafi, 171-187. NEEDHAM, S. & MACKLIN, M. G. (eds) 1992. Alluvial Archaeology in Britain. Oxbow Monograph, 27, Oxford. NESJE, A. & JOHANNESSEN, T. 1992. What were the primary forcing mechanisms of highfrequency Holocene glacier and climatic variations. The Holocene, 2, 70-84. PATZELT, G. 1985. The period of glacier advances in the Alps, 1965 to 1980. Zeitschriftffir Gletscherkunde und Glazial-geologie, 21,403-407. ROTNICKI, K. 1991. Retrodiction of palaeodischarges of meandering and sinuous alluvial rivers and its palaeohydroclimatic implications. In: STARKEL L., GREGORY, K. J. & THORNES, (eds) Temperate Palaeohydrology. J. Wiley, 431-471. RUDOY, A. N. & BAKER, V. R. 1993. Sedimentary effects of cataclysmic late Pleistocene glacial outburst flooding, Altay Mountains, Siberia Sedimentary Geology, 85, 53-63. SCHWETZ,G. I. 1978. Multi-centenial changes of the run-off of the Dniepr river [in Russian], Gidrometeoizdat, Leningrad. SHIKLOMANOV,I. A. 1990. Global water resources. Nature and Resources, 26, 3, 34-43. SOJA,R. 1994. Paleohydrologia ilo~ciowa [Quantitative palaeohydrology] [in Polish]. Przeglqd Geograficzny, T.L.XVI, 1-2. SONNTAG, C., KLITZCH, E. & LOEHNERT, E. P. 1980. Isotopic identification of Saharian groundwater; groundwater formation in the past. Palaeoecology of Africa, 12, 159-171. STARKEL, L. 1972. The role of catastrohic rainfall in the shaping of the relief of the lower Himalaya (Darjeeling Hills). Georgraphica Polonica, 21, 103-147. - - 1 9 7 6 . The role of extreme (catastrophic) meteorological events in the contemporary evolution of slopes. In: DERBYSHIRE, E. (ed.) Geomorphology and Climate. J. Wiley, 203-246. -
-
-
-
PALAEOHYDROLOGICAL RECONSTRUCTION
17
- - 1 9 8 3 . The reflection of hydrologic changes in the fluvial environment of the temperate zone during the last 15000 years, In: GREGORY, K. J. (ed.) Background to Palaeohydrology.'A Perspective. J. Wiley, 213-235. - - 1 9 8 5 . The reflection of the Holocene climatic variations in the slope and fluvial deposits and forms in the European mountains. Ecologia Mediterranea, 11, 91-98. - - 1 9 9 0 . Fluvial environment as an expression of geoecological changes. Zeitschrift fur Geomorphologie, Suppl. 79, 133-152. - - 1 9 9 1 a, Long-distance correlation of fluvial events in the temperate zone. In: STARKEL, L., GREGORY, K. J. & THORNES, (eds) Temperate Palaeohydrology. J. Wiley, 473-493. - - 1 9 9 1 b . The Vistula river valley: a case study for Central Europe. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology. J. Wiley, 171-188. - - 1 9 9 1 . Late Quaternary continental palaeohydrology as related to future environmental change. Global and Planetary Change, 7, 95-108. --, Mass movements during the Holocene: the Carpathian example and the European perspective. In: FRENZEL, B. et al. ESF workshop on rapid mass movements, Mainz. Palaeoklimaforschung, in press. STOCKTON, C. W., BOGESSS, W. R. & MEKO, D. M. 1985. Climate and Tree rings. In: HECHT, A. D. (ed.) Paleoclimatic Analysis and Modelling, J. Wiley. STREET-PERROTT, F. m. & HARRISON, S. P. 1985. Lake levels and climate reconstruction. In: HECHT, A. D. (ed.) Paleoclimate Analysis and Modelling, J. Wiley, 291-340. SWAIN, A. M., KUTZBACH, J. E. & HASTENRATH, S. 1983. Estimates of Holocene precipitation for Rajasthan, India, based on pollen and lake-level data. Quaternary Research, 19, 1-17. WEBB, T., III 1983. Calibration of Holocene pollen data in climatic terms. Quaternary Studies in Poland, Poznafi, 4, 107-113. & BRYSON, R. A. 1972. Late and Postglacial climatic change in the Northern Midwest, USA: Quantitative estimates derived from fossil pollen spectra by multivariate statistical analysis. Quaternary Research, 2, 70-115. -
-
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 19-25
Palaeohydrology and future climate change NIGEL
W. ARNELL
Department of Geography, University of Southampton, Southampton S0171B J, UK Since the mid-1980s there has been growing concern that increasing concentrations of so-called greenhouse gases will lead to global warming, changes to regional climates, and hence impacts on the environment, society and economy. The United Nations Environment Program and the World Meteorological Organization in 1987 set up the Intergovernmental Panel on Climate Change (IPCC), which has reported on climate change predictions (IPCC 1990a, 1992a), the possible impacts of climate change (IPCC 1990b, 1992b) and strategies for mitigating the effects of climate change (IPCC 1991). The IPCC Second Assessment is scheduled for publication in 1996. Climate change science is essentially predictive: it is trying to predict conditions during the twenty-first century. By far the most common approach is based around the use of global climate models (GCMs). These numerical simulation models are used to predict the global and regional climatic effects of changing greenhouse gas concentrations, and many climate change impact studies use scenarios based in some way on GCM simulations. These scenarios are then used to perturb current climatic time series, and fed through a catchment hydrological model to simulate river flows and other hydrological properties (e.g. Bultot et al, 1988; Lettenmaier & Gan 1990; Arnell & Reynard 1993, and many others). The two major problems with this approach lie in the definition of credible catchment-scale scenarios from GCM simulations, and the development of realistic hydrological simulation models. The latter problem is fundamental to hydrological simulation, whilst the former is peculiar to climate change impact assessments and arises for two reasons. First, GCMs do not at present represent all the climatic processes in a realistic manner, particularly those relating to the development of clouds and the interactions between the atmosphere and the land surface, and second GCMs operate at a very coarse spatial resolution. Some important atmospheric processes, such as the development of meso-scale circulation patterns and convective storms, are therefore not simulated particularly well, and whilst GCMs can simulate large-scale atmospheric features well, regional and local climates are often not well reproduced. The coarse spatial resolution also means that GCM output has to be interpolated down to the catchment scale. A variety of techniques of varying degrees of sophistication have been used or proposed, ranging from simple statistical interpolation through empirical relationships between large-scale and local climate to the use of nested regional climate simulation models, but all rely ultimately on the reliability of the GCM simulations of large-scale climatic features (Arnell 1995a). The other popular approach to climate impact assessment uses the past as an analogue for the future. One variant uses the instrumental period, in practice the last century, another uses historical data, and a third uses palaeoclimatic reconstructions.
20
N . W . ARNELL
This paper considers the potential contribution of palaeohydrology to the prediction of future climates and hydrogeomorphological characteristics, focusing on two issues: the contribution of palaeohydrological reconstructions to climate model validation, and the use of palaeohydrological analogues.
Climate modelling and palaeohydrological reconstructions Global climate models (GCMs) are numerical simulation models, based on the laws of physics and applying the principles of conservation of mass and energy (for a comprehensive review see Henderson-Sellers 1994). They operate on a grid network, with the highest resolution global-scale models currently working on a grid resolution of the order of 250 • 250 km. Many significant atmospheric processes, including cloud formation at the micro-scale and the development of depressions at the meso-scale, operate at a finer spatial resolution than can be resolved by current GCMs, so are modelled through simplified parameterized representations. GCMs are used for climate prediction and the development of climate change scenarios, and the credibility of such predictions and subsequent impact assessments is dependent on the credibility of the climate model. It is of course not possible to validate predictions of future climate change, but it is possible to validate simulations of current climate. However, a model that simulates current climate well does not necessarily produce a good simulation of future climate. Fortunately, it is possible to assess the ability of a GCM to simulate changed climates using palaeoclimatic and palaeohydrologic data (Street-Perrott & Roberts 1994). The CLIMAP and COHMAP projects assembled comprehensive palaeoclimatic and palaeo-vegetation data sets for the period since the last glacial maximum (18kaBP: CLIMAP Project Members 1976; COHMAP Members 1988), and these data have been used both to provide the boundary conditions for GCM experiments and to validate model simulations (IPCC 1990a). There is little information on temperature and precipitation for these early periods, so model validation has largely been based on a comparison of palaeo-vegetation (reconstructed from pollen data) and 'synthetic' vegetation, predicted using current relationships between indices of vegetation and temperature and precipitation. In general, the palaeoclimatic modelling studies have found good agreement between simulations and observations at continental scales (IPCC 1990a), although there are regional differences. The Palaeoclimate Modelling Intercomparison Project (PMIP), part of the IGBP PAGES (Past Global Changes) project, is currently attempting to validate climate model simulations for three periods - 6 ka BP, 18 ka BP and 115 ka BP - using a variety of palaeoclimatic data. The time periods being used were selected both because they represented periods with different boundary conditions (insolation changes at 6 kaBP and the presence of ice sheets at 18 kaBP, for example) and because data were available. The US National Science Foundation project TEMPO (Testing Earth System Models with Palaeoenvironmental Observations) is contributing to PMIP. In principle, palaeohydrological data can be used for the quantitative evaluation of GCM simulations, although there are a number of complications with using river runoff data for model validation and very few attempts have been made using c u r r e n t data (Arnell 1995b). One approach that has been adopted is 'top-down', using data from large river basins covering several climate model grid cells (Russell
PALAEOHYDROLOGY AND FUTURE CLIMATE CHANGE
21
& Miller 1990). A 'bottom-up' approach would try to build up information characterizing each individual climate model grid cell from small and medium-sized catchments within that cell (Arnell 1995b). Palaeohydrological reconstructions have the potential to assist in climate model validation /f it is possible to reconstruct seasonal or annual runoff totals. Reconstructed lake levels, particularly for lakes in closed basins, are another source of quantitative data on the balance between precipitation, evaporation and runoff (Kutzbach & Street-Perrot 1985; Fontes & Gasse 1991; Harrison & Digerfeldt 1993), and been widely used in climate model validation.
Palaeohydrologic reconstructions as analogues for future climate In principle, information from the geological past could provide useful insights into the dynamics of the natural environment and links between climate and hydrogeomorphological response, and could provide analogues for future climatic conditions. Three palaeoclimatic analogues have been identified by the Russian school (IPCC 1990a): (1) (2) (3)
the Holocene climatic optimum (6.2-5.3 kasP), representing an increase in temperature of I~ the last interglacial (125 ka BP), representing a 2~ increase in temperature; the Pliocene (3 to 4 Ma BP), representing a rise of 4~
However, there are a number of theoretical and practical problems with using such analogues. There are three major theoretical assumptions. First, it is assumed that the relationships between form and process operating today are the same as those operating in the p a s t - the uniformitarian assumption- and that it is possible to infer past process from past form. Second, it is assumed that the data from the past represent equilibrium conditions and this may not be true in practice; much palaeohydrological evidence reflects periods of adjustment to altered conditions. Third, and most importantly, it is assumed that the local effects of a change in climate are independent of the causes of the change in climate. Variations in global climate over geological time scales, however, are a result of changes in the Earth's orbit, and hence changes in the spatial pattern of receipt of solar radiation. Mitchell (1990) showed, for example, that the changes in net radiative forcing during the Holocene optimum were very different to the changes expected under a rising concentration of greenhouse gases; the spatial variability in forcing during the Holocene was considerably greater. Not only are such forcing conditions different, but boundary conditions, especially the extent of ice cover, have varied over geological time, affecting global and regional climates. The effects of such variations in forcing factors and boundary conditions over time may be rather different to the effects of a greenhouse gas-induced forcing, and palaeoclimates do not provide good analogues for future climate change (IPCC 1990a). The major practical problem with the use of palaeohydrological data as an analogue lies in the derivation of quantitative information at relevant time and space scales: in other words, inferring process from form. Different techniques are used for
22
N.W. ARNELL
different components of the water balance, and many are illustrated in this volume. Precipitation can be estimated from past vegetation patterns, derived through the analysis of pollen extracted from dated lake and peat bog cores, and from tree ring analysis. Multivariate relationships between vegetation pattern, rainfall and temperature, for example, have been calibrated using current data and applied to palaeo-vegetation patterns to infer both precipitation and temperature (COHMAP Project Members 1988). There are many examples of statistical relationships between tree ring width and precipitation, again calibrated on current climatic data (e.g Till & Guiot 1990; Cleveland & Duvick 1992), although these tend to extend back in time for at most 1000 years. River discharge can be inferred from palaeochannel form and sediment analysis. Inference from form is based on the use of hydraulic equations, predicting flows from channel dimensions and planform characteristics (Maizels 1983; Gregory & Maizels 1991; Rotnicki 1991). These equations, however, require information which cannot readily be determined from geological evidence such as channel roughness, and it is difficult to draw inferences about hydrological regimes just from data about channel-forming discharges. Sedimentary data are most useful for reconstructing peak flows, again based on current empirical relationships between discharge and sediment deposition patterns. Knox (1993), for example, reconstructed a 7000-year flood chronology for upper Mississippi river tributaries from dated floodplain gravel deposits. He showed that a small change in climate could result in a large change in flood magnitudes. River flows have been estimated from tree rings (e.g. Earle 1993; Loaiciga et al. 1993; Young 1994), but this relies on intermediate relationships with precipitation and, as with reconstructions of precipitation, it is difficult to go back even as far as 1000 years. Records of lake storage are rather easier to construct from both relict erosional features and from shoreline deposits (e.g. Magny 1992), and lake-level changes can give information on changes in the balance of precipitation and evaporation. Groundwater storage at different periods in the past can be estimated through the dating of groundwater (Love et al. 1994), and also in some cases by geological and geochemical evidence left behind by higher water tables. It is also possible to model changes in groundwater storage, using reconstructed climate data, validating the model results against dated groundwater or groundwater chemical characteristics (Hiscock & Lloyd 1992). Whilst it may be possible to derive some quantitative information on past changes in hydrological regime, it is difficult to see how this can be used as an analogue for future climates; the factors causing change are different, both the forcing factors and boundary conditions are different, and the temporal resolution of the data will be low. Many of the events recorded in the geological record reflect changes from glacial to inter-glacial conditions, not changes from one non-glacial state to another.
Understanding landscape and climate dynamics: the contribution of palaeohydrological research to future climate change studies Although palaeohydrological research does not contribute directly to research into future climate change, it has a major potential role in the development of improved understanding of both the dynamics of climate over different time scales and of the
PALAEOHYDROLOGY AND FUTURE CLIMATE CHANGE
23
response of the fluvial system to change. The Past Global Changes (PAGES) project of the IGBP is aimed at providing a quantitative understanding of the earth's past environments, as a guide to defining the envelope of natural variability (Eddy 1992), and palaeohydrological research is clearly highly relevant to this aim. Over the last few years it has become increasingly accepted that climate time series in many parts of the world show clear non-random behaviour, and that there are strong correlations in behaviour, or teleconnections, between different parts of the world. The E1 Nifio/Southern Oscillation phenomenon, for example, affects circulation patterns, rainfall and river runoff every few years across much of the southern hemisphere (Kuhnel et al. 1990; Mechoso & Iribarren 1992; Simpson et al. 1993), and its signal can be seen in river flows in the western United States (Aguado et al. 1992; Kahya & Dracup, 1994). The sub-discipline of hydroclimatology is developing strongly, seeking explanations for such patterns and linkages and attempting to use these explanations to refine understanding of climate dynamics; with an improved understanding of such dynamics and correlations it should be possible to make more credible predictions for the future. Within Europe, one of the aims of the F R I E N D project (Flow Regimes from International Experimental and Network Data: Gustard 1993; Arnell 1994) is to investigate links between atmospheric behaviour and large-scale hydrological anomalies, using just the observed river flow record. Palaeoclimatic and palaeohydrological reconstructions can provide valuable longterm time series for hydroclimatological studies; Wells (1990), for example, has reconstructed the history of the E1 Nifio phenomenon in Peru through the Holocene from flood sediments. Palaeohydrological reconstructions provide information on how the fluvial system responded to change. Even though these changes, such as de-glaciation and recovery from a glacial period, are not directly relevant to future global warming due to increasing greenhouse gas concentrations, the information provided can be used to test models of change and hypotheses about landscape dynamics. Validated models of change are essential for credible predictions of the future.
References AGUADO, E., CAYAN, D., RIDDLE, L. & Roos, M. 1992. Climatic fluctuations and the timing of West Coast streamflow. Journal of Climate 5, 1468-1483. ARNELL, N. W. 1994. Variations over time in European hydrological behaviour: a spatial perspective. In: FRIEND: Flow Regimes from International Experimental and Network Data. IAHS Publications 221, 179-184. - - 1 9 9 5 a . Scenarios for hydrological climate change impact studies. In: OLIVER, H. R. & OLIVER, S. A. (eds) The Role of Water and the Hydrological Cycle in Global Change. Springer, Berlin, 389-407. - - 1 9 9 5 b . River runoff data for the validation of climate simulation models. In: OLIVER, H. R. & OLIVER, S. A. (eds) The Role of Water and the Hydrological Cycle in Global Change. Springer, Berlin, 349-371. -& REYNARD, N. S. 1993. Effect of climate change on river flow regimes in the United Kingdom. Institute of Hydrology. Report to Department of the Environment. BULTOT, F., COPPENS, A., DUPRIEZ, G. L., GELLENS, D. & MEULENBERGHS, F. 1988. Repercussions of a CO2 doubling on the water cycle and on the water balance - a case study for Belgium. Journal of Hydrology, 99, 319-347. CAYAN, D. R., RIDDLE, L. G. & AGUADO, E. 1993. The influence of precipitation and temperature on seasonal streamflow in California. Water Resources Research, 29, 1127-1140.
24
N.W. ARNELL
CLEVELAND, M. K. & DUVICK, D. N. 1992. Iowa climate reconstructed from tree rings, 1640-1982. Water Resources Research 28, 2607-2615. CLIMAP PROJECT MEMBERS 1976 The surface of the ice-age Earth. Science, 191, 1131-1136. COHMAP MEMBERS 1988. Climatic changes of the last 18,000 years: observations and model simulations. Science 241, 1043-1052. EARLE, C. J. 1993. Asynchronous droughts in California streamflow as reconstructed from tree rings. Quaternary Research, 39, 290-299. EDDY, J. A. (ed.) 1992. PAGES: Past Global Changes Project. Proposed Implementation Plans for Research Activities. IGBP Global Change Report No. 19. Stockholm, Sweden. FONTES, J. C. & GASSE, F. 1991. PALHYDAF (Palaeohydrology in Africa) p r o g r a m objectives, methods, major results. Palaeogeography, Palaeoclimatology, Palaeoecology, 84, 191-215. GREGORY, K. J, & MAZIELS, J. K. 1991. Morphology and sediments: typological characteristics of fluvial forms and deposits. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15000 Years. Wiley, Chichester, 31-59. GUSTARD, A. (ed.) 1993. Flow Regimes from International Experimental and Network Data (FRIEND). Institute of Hydrology, Wallingford, Oxon. HARRISON, S. P. & DIGERFELDT, G. 1993. European lakes as palaeohydrological and palaoeclimatic indicators. Quaternary Science Reviews, 12, 233-248. HENDERSON-SELLERS, A. 1994. Numerical modelling of global climates. In: ROBERT, N. (ed.) The Changing Global Environment. Blackwell, Oxford 99-124. HISCOCK, K. M. & LLOYD, J. W. 1992. Palaeohydrogeological reconstructions of the North Lincolnshire Chalk, UK, for the last 140000 years. Journal of Hydrology, 133, 313-342. IPCC 1990a. Climate Change: the IPCC Scientific Assessment. Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. - - 1 9 9 0 b . Climate Change: the IPCC Impacts Assessment. Intergovernmental Panel on Climate Change. Australian Government Publishing Service, Canberra. - - 1 9 9 1 . Climate Change: the IPCC Response Strategies. Intergovernmental Panel on Climate Change. Island Press, Washington DC. - - 1 9 9 2 a . Climate Change 1992: The Supplementary Report to the 1PCC Scientific Assessment. Intergovernmental Panel on Climate Change. Cambridge University Press, Cambridge. - - 1 9 9 2 b . Climate Change 1992: The Supplementary Report to the IPCC Impacts Assessment. Intergovernmental Panel on Climate Change. World Meteorological Organization and United Nations Environment Program. KAHYA, E. & DRACUP, J. A. 1994. The influences of Type 1 E1 Nifio and La Nina events on streamflows in the Pacific Southwest of the United States. Journal of Climate, 7, 965-976. KNOX, J. C. 1993. Large increases in flood magnitude in response to modest changes in climate. Nature, 361,430-432. KUHNEL, U., MCMAHON, T. A., FINLAYSON, B. L., HAINES, A., WHETTON, P. H. & GIBSON, T. T. 1990. Climatic influences on streamflow variability: a comparison between southwestern Australia and southeastern United States of America. Water Resources Research, 26, 2483-2496. KUTZBACH, J. E. & STREET-PERROTT, F. m. 1985. Milankovitch forcing of fluctuations in the level of tropical lakes from 18 to 0 kyr BP. Nature, 317, 130-134. LETTENMAIER, D. P. & GAN, T. Y. 1990. Hydrologic sensitivities of the Sacremento-San Joaquin River basin, California, to global warming. Water Resources Research, 26, 69-86. LOAICIGA, H. A., HASTON, L. & MICHAELSEN, J. 1993. Dendrohydrology and long-term hydrologic phenomena. Reviews of Geophysics, 31, 151-171. LOVE, A. J., HERCZEG, m. L., LEANLY, F. W., STADTER, M. F., DIGHTON, J. C & ARMSTRONG, D. 1994. Groundwater residence time and palaeohydrology in the Otway Basin, South Australia - H2, O 18 and C 14 data. Journal of Hydrology, 153, 157-187. MAGNY, M. 1992. Holocene lake level fluctuations in Jura and the northern sub-Alpine ranges, France - regional pattern and climatic implications. Boreas, 21, 319-334.
PALAEOHYDROLOGY AND FUTURE CLIMATE CHANGE
25
MAZIELS, J. K. 1983. Palaeovelocity and palaeodischarge determination for coarse gravel deposits. In: GREGORY, K. J. (ed.) Background to Palaeohydrology. Wiley, Chichester, 101-139. MECHOSO, C. R. & IRIBARREN, G. P. 1992. Streamflow in southeastern South America and the Southern Oscillation. Journal of Climate 5, 1535-1539. MITCHELL, J. F. B. 1990. Greenhouse warming: is the mid-Holocene a good analogue? Journal of Climate, 3, 1177-1192. ROTNICKI, K. 1991. Retrodiction of palaeodischarges of meandering and sinuous alluvial rivers and its palaeohydroclimatic implications. In: STARKEL, L., GREGORY, K. J. & THORNES, J. B. (eds) Temperate Palaeohydrology: Fluvial Processes in the Temperate Zone during the last 15000 Years. Wiley, Chichester, 431-471. RUSSELL, G. L. & MILLER, J. R. 1990. Global river runoff calculated from a global atmospheric general circulation model. Journal of Hydrology, 117, 241-254. SIMPSON, H. J., CANE, M. A., LIN, S. K., ZEBIAK, S. E. & HERCZEG, A. L. 1993. Forecasting annual discharge of the River Murray, Australia, from a geophysical model of ENSO. Journal of Climate, 6, 386-390. STREET-PERROTT, F. A. & ROBERTS, N. 1994. Past climates and future greenhouse warming. In: ROBERTS, N. (ed.) The Changing Global Environment. Blackwell, Oxford, 48-68. TILL, C. & GUIOT, J. 1990. Reconstruction of precipitation in Morocco since 1100 AD based on cedrus atlantica tree ring widths. Quaternary Research, 33, 337-351. WELLS, L. E. 1990. Holocene history of the E1 Nifio phenomenon as recorded in flood sediments of northern coastal Peru. Geology, 18, 1134-1137. YOUNG, K. C. 1994. Reconstructing streamflow time series in central Arizona using monthly precipitation and tree ring records. Journal of Climate, 7, 361-374.
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 27-42
Changes in moisture balance between glacial and interglacial conditions: influence on carbon cycle processes J. M . A D A M S
1'2 & H . F A U R E 2
1School of Geography, University of Oxford, 1 Mansfield Road, Oxford OX13TB, UK. Present address." Kings" College, Cambridge CB21ST, UK 2 Laboratoire de Geologie du Quaternaire, CEREGE, Europole de l'Arbois, BP 80, F-13545, Aix-en-Provence Cedex 04, France. Abstract: During the arid Late Glacial and Last Glacial Maximum (between
approximately 30 000 and 13 000 calendar years ago), vegetation cover retreated and large areas of the continents were occupied by desert and semi-desert vegetation. The result of this general decrease in biological activity would have been a decrease in the size of the land carbon reservoir, and a decrease in the rate of chemical rock weathering. By contrast, during the early-mid-Holocene, conditions in many areas seem to have been moister than today due to a more active hydrological cycle. All of these processes would have affected the global carbon cycle and altered the amplitude and timing of the climate fluctuations themselves. In effect, the climatic shift between glacial and interglacial conditions creates a very large 'missing source' of carbon, perhaps amounting to thousands of gigatonnes, to account for the carbon uptake by the land system during the present interglacial, and thus carbon cycle models of the late Quaternary may need to be revised extensively.
Evidence from polar ice cores indicates that the levels of the greenhouse gases CO2 and CH4 in the Earth's atmosphere have closely paralleled the changing climate over at least the last 250000 years. (e.g. Neftel et al. 1988; Barnola et al. 1989; Lorius et al. 1990; Chapellaz et al. 1993; Alley et al. 1993) (Fig.l). There is currently widespread interest in the controls on these changes in atmospheric composition, and global hydrological processes seem likely to play a dominant role in explaining how and why the carbon cycle changed at this timescale. In this paper the influence of these hydrological processes on CO2 fluctuations is discussed; this influence is considered to be quantitatively much more important than CH4 in reinforcing glacialinterglacial climate differences. There can be little doubt that on the glacial-interglacial timescale the dissolved ocean carbon reservoir dominates the system, because it contains many times more carbon than is present in land ecosystems (40 000 Gt as opposed to around 2000 Gt; Bolin et al. 1977). Understandably, therefore, effort has mainly concentrated on producing ocean models (Broecker & Peng 1993) to explain the sequestering of larger amounts of carbon into the glacial-age ocean. More recently, however, there have been various attempts to provide a more complete picture by estimating how land carbon storage might have changed between glacial and interglacial conditions
28
J. M. A D A M S & H. F A U R E
(b)
(a)
AT
500
~
Depth (m) 1000 1500
AT
2000
"
!
2 0 -2 -4 -6 -8 -10
2. C
Depth (m) 1000 1500
500
2000
*C 2~
o~
o -2
-2
-4
-4
-6
-6
-8
-8
-10
CO= ppmv 300 280 260
AT
CN, (ppbv) 700
-10
CO= ~prnv 300 28O
600
500
260 240
240 220 200
180
180 o
' 4b
,
'
8'o
'
,~o
'
Age (kyr BP)
l
40
,
j
,
80 Age (kyr BP)
,
120
~
400
300 160
Fig. 1. (a) History of atmospheric carbon dioxide levels reconstructed from the Vostok Antarctic ice core record (Barnola et al. 1987). (b) History of atmospheric methane levels reconstructed from the Vostok Antarctic ice core record (Chapellaz et al. 1990).
(e.g. Faure 1989; Adams et al. 1990; van Campo et al. 1993; Peng 1994; Crowley 1995a, b; Adams 1995a, b). Such studies based on palaeoenvironmental reconstruction have suggested that the difference in water balance between full-glacial and fullinterglacial conditions was greater than had previously been thought. With this large contrast in the hydrological balance of the continents, combined with major temperature changes, there would have been large changes in their role in the global carbon cycle and a feedback on the climate system itself. Here, the discussion concentrates on the picture suggested by recent mapping efforts to reconstruct the world's land vegetation cover based on fossil and palaeoclimatic evidence.
Evidence for changes in global water balance Following the earlier view that glacial periods were always cooler and moister than at present, by the early 1950s it became apparent that the Last Glacial or oxygen Isotope Stage 2 (a period extending between about 30 000 and 13 000 years BP), and in particular the Last Glacial Maximum (LGM, about 20 000 years BP), was more arid than present throughout most of the world. As more evidence from the LGM has been presented, the extent and severity of this stage of aridity has become more strikingly evident. In contrast, evidence for moister-than-present conditions during the early-mid-Holocene has also been found for many parts of the world. There have been a number of attempts to bring together evidence that relate, directly or indirectly, to changes in continental moisture balance over this timespan. The history of changes in water level in lake basins is a major source of data as a palaeohydrological indicator (Street-Perrott & Roberts 1983; Wright 1993).
MOISTURE CHANGES & CARBON CYCLE
29
Additionally, pollen databases such as that of Webb et al. (1995) use the plant fossil remains of past vegetation cover, which can be used to reconstruct the general climatic conditions of the past. The baselines for general palaeoclimatic 'atlases' are being established (Wright 1993), but the process is in its early stages for many regions because of the shortage of data than can be readily translated into a quantitative form. There have been several recent attempts to reconstruct palaeovegetation on a global scale, using fossil and/or sedimentological evidence. These include maps by van Campo et al. (1993), Adams (1995b) and Crowley (1995a). Such palaeovegetation maps can be regarded as an indirect summary of the hydro-climatological regime of the past; for instance closed forest vegetation (defined in terms of a forest canopy cover formed by the tree crowns being close enough to touch one another) indicates moister conditions than desert or semi-desert. A recent attempt to collate the range of these and other sources of data into a coherent picture of biome distribution has been the Quaternary Environments Network (Adams 1995a, b; QEN 1995; Adams & Faure in press) review of palaeovegetation cover, based on a range of sources of plant fossil, sedimentological and zoological evidence which relate directly or indirectly to vegetation cover. The QEN biome map reconstructions for the LGM, the early Holocene and the present-natural illustrate the extent of change in the water balance, with the moist-climate tropical and temperate rainforest area being much reduced at the LGM (indicating aridity), and expanded during the early Holocene (due to more humid climates). Global deserts show the opposite pattern, being more extensive during the LGM and less extensive than present during the early-mid-Holocene. The Sahara desert almost disappeared under a vegetation cover during the early Holocene, indicating a moister climate than at present, and the same seems to have been true of the central Asian deserts. GCM modelling experiments, using prescribed sea surface temperatures and ice distributions, also suggest that the LGM world was substantially drier than at present, and that the early-mid-Holocene would on average have been moister due to the Milankovitch effect of greater summer radiation on the northern continents (Prentice et al. 1993; Wright 1993). There is a disparity of evidence in that the computer model simulations for the LGM predict conditions substantially moister than the palaeoevidence would appear to suggest. However, such computer simulations are still at an early stage of development, with heat flux corrections continually necessary to make them conform to 'realistic' climate patterns (Crowley & North 1991; Kagan 1995). The boundary conditions against which the models are established and continually returned to may also turn out to be incorrect. For example, the prescribed sea surface temperatures (e.g. CLIMAP 1976) that published model studies have always used are now thought to be too warm for the LGM of the tropics (Crowley 1994; Broecker 1995). If LGM sea surface temperatures used in the models were reduced, the result would be even drier conditions over the continents, bringing the results more closely into line with palaeoevidence. However at present there is too much uncertainty regarding the reliability of GCM models to use them to reconstruct global changes in the past hydrological regime. For this reason we concentrate here on the picture gathered from more direct forms of palaeoevidence (Starkel 1989). When discussing past vegetation cover and ecosystem processes in relation to past hydrological balance, it is necessary to recognise that the observed changes may not
30
J. M. ADAMS & H. FAURE
be the result of real climatic changes in rainfall or potential evapotranspiration. This is because the hydrological balance of plants and vegetation is also partly controlled by the ambient CO2 concentration: the less CO2 there is available, the more that plants need open their stomata (the pores in their leaves) to obtain it (Woodward 1987). The more that plants open their stomata, the more moisture they lose. Thus, lower CO2 levels alone could mimic the effects of a 'drier' climate, causing vegetation zones to shift their positions along climatic gradients. However, the true importance of the relationship between CO2 changes and vegetation ecology in the past remains highly uncertain, despite a great deal of speculation. Lake level evidence suggests that there was much lower rainfall at the LGM in most areas, relatively independent of vegetation processes (Street-Perrott & Roberts 1983; Wright 1993). Closed chamber experiments studying the effects of raised 'future' CO2 levels on plants give confusing and sometimes contradictory results (McConnaughay et al. 1993; Koerner & Arnone 1993; Mooney & Koch 1994), and there is no clear evidence that the rapid 80-90 ppm CO2 rise over the last 200 years has had any effect in increasing vegetation cover and productivity (Adams & Woodward 1992; Shiel & Philips 1995). It seems reasonable to conclude at present that most of the aridity of the LGM world was due to a reduction in rainfall and not 'apparent' aridity due to a direct effect of CO2.
Changes in bulk carbon storage For practical purposes of study, the world's vegetation cover of 'biomes' can be used to define around 35 'ecosystems', including not only the vegetation of each biome but also the litter, soils and peat layers underneath. Carbon storage in the components of these land ecosystems is closely related to local and regional hydrological conditions. Thus, in almost any given region of the Earth's surface, there is a striking broadscale relationship between annual precipitation and natural or semi-natural vegetation carbon storage. The relationships are even closer if differences in temperature are considered as a factor affecting moisture availability. Forest ecosystems, the richest in terms of vegetation carbon storage (Olson et al. 1983; Harmon & Hua 1992), require a greater moisture supply than dry grassland ecosystems, and if there is no moisture available, there is a desert with no vegetation carbon store. In a study of the Mediterranean region, Le Houerou & Hoste (1977) report an exponential relationship between vegetation biomass and annual precipitation. Even within individual biomes/ecosystems, such as the steppe biome, there is often a strong relationship between annual rainfall and vegetation carbon storage (Le Houerou & Hoste 1977; Olson et al. 1983). Soil carbon shows a similar positive relationship with moisture availability (the balance between precipitation and evaporation), illustrated in the summary diagrams of Fig. 2 and Post et al. (1982). This is because the rate of input of dead plant parts from the vegetation to the soil tends to be greater under moist conditions, other factors being equal. Also, above a certain level of moisture content the soil is too waterlogged to allow efficient aerobic decay of plant material, thus promoting peat formation. Given such simple, general relationships between moisture availability and ecosystem carbon storage, it is evident that a change in the moisture regime, whether due to real climatic changes or to a direct-CO2 effect on plant water balance, would affect global carbon storage on land. With respect to carbon storage, attention
31
MOISTURE CHANGES & CARBON CYCLE
Polar
Subtropi Tropical cal
l /--~
~
~ /__~k_
- i - r. -, -~'~lrdB,s
- .~4,.- 7 -
Warmtemperate
500 S I ','..'.'.,'..\ .
X~ 1 / /,-~____ '..~.~=.==~163
s162 #
r
-
\-
~
-
' ; _ ~ _ - \s ~ - ~ ' ~ ' ~ . " ~ _ ~ . " ~
~
~/- "~~~ / ~ ~ ~ 2
f~ ~ ~uJs 46 8
10
Carbon in mineral soil (kg
. . .
Nlval
1.5~
%
~
"
~Ll~a~n~~
- -"
3~
m
~8000 L~ m~ Pre~ntane
24*
14 18 22 m-2)
Fig. 2. Summary of the relationship between climate and soil carbon storage found in the database of Zinke et al. (1984). Reprinted with permission from Nature. (Post et al., vol. 298, p. I56). Copyright (1982) Macmillan Magazines Ltd. has focused on the LGM as the most 'extreme' part of the Stage 2 period in terms of aridity, but the same general picture is also relevant for much of the time period, particularly after c.25 000 BP. There are a number of methods to estimate the change in carbon storage between the opposite climatic extremes of the LGM and the earlymid-Holocene. These methods are in three different categories; (i) reconstruction of palaeovegetation/ecosystem cover for glacial and interglacial conditions (e.g, Adams et al. 1990; Crowley 1995a, b), (ii) use of general circulation models (GCMs) to 'predict' past climates and using these data to 'predict' palaeovegetation cover and carbon storage (e.g. Prentice et al. 1993), (iii) taking the magnitude of the carbon isotope change in the oceans between glacial and interglacial conditions as an indicator of the size of the land carbon storage change (e.g. Shackleton 1977). All three methods suggest that there was a net increase in land carbon storage of at least several hundred gigatonnes between the LGM and the mid-Holocene (Fig. 3) (Maslin et al. 1995; Adams 1995a, b). The ocean-isotope method gives the lowest estimates (350-700 Gt) (Shackleton 1977; Currey et al. 1988; Broecker & Peng 1993) with palaeoevidence-based estimates tending to be larger at around 700-1700 Gt, with GCM-based studies giving intermediate values (Freidlingstein et al. 1992; Prentice et al. 1993). Each method of estimating total land carbon storage has inherent sources of error (Maslin et al. 1995; Crowley 1995a, b), and it appears that the interdisciplinary palaeoevidence-based approach is likely to be the most robust as it offers the most direct indication of the changing nature of land ecosystems themselves. Of the palaeoevidence-based studies (Fig. 3), several have produced estimates of a relatively low shift in carbon storage (400-1000 Gt C) due to (i) them incorporating a narrow range of palaeoenvironmental evidence, (ii) the assignment of present-day low anthropogenic carbon storage values to essentially pre-anthropogenic Holocene forest
GCM
I
i
GCM
Freidlingstein et al. (1992) 300-600 Gt
I
I
13C
I----I Currey et al. (1988) 650 Gt
13C
13C
I
Broecker & Peng (1993) 425 Gt
Adams & Faure (in press) 1000-2300 Gt
Palaeoevid.
Palaeoevid.
I
I
I
250
I
500
Crowley(1995 a,b) 750-1050 Gt Peng. etal.(in press) 469-950 Gt
van Campo etal.(1993) 400-700 Gt
Palaeoevid.
0
I
[
Palaeoevid. I
I
Shackleton (1977) 500-1000 Gt
I
t~l
Palaeoevid.
Prentice et al. (in press) 400-900 Gt
Adams et al. (1990)
I
750
I
1000
H
I
I
1250
1500
I
1750
2000
LGM-to-lnterglacial increase in land carbon (Gt) Fig. 3. Estimates for the LGM-Holocene increase in organic carbon in vegetation, soils and peatlands on land. After Maslin et al. (1995) Adams (1995b), gathered from sources not directly cited here. Palaeoevid, palaeoevidence-based reconstruction of ecosystems; GCM, climate model-based prediction of past ecosystems; ~3C, estimate based on shifts in the ocean 13C/12Cratio.
MOISTURE CHANGES & CARBON CYCLE
33
ecosystems, and (iii) the assumption that LGM desert and semi-desert ecosystems were as rich in carbon as present-day moist steppe. Given the bulk of evidence of global LGM aridity and carbon-poor soils, it seems appropriate to concentrate here on scenarios in which a relatively large (approximately 1000-1700 Gt) shift in land carbon storage occurs, though without neglecting the scenarios in which a somewhat smaller shift (around 500 Gt) occurs. If one takes a conservative value of around 1000GtC as the amount of carbon released from land ecosystems as the Earth shifts from interglacial to full glacial conditions, this would release 476ppm of CO2 into the atmosphere (if 2.1 Gt C ,,~ 1 ppm CO2). On the timescale of thousands of years this CO2 would distribute itself between the atmosphere and the ocean, with the largest fraction of the CO2 being dissolved into the ocean water as CO2 gas and as bicarbonate. According to the alkalinity buffer system (Siegenthaler 1989), around 16% of the extra CO2 would stay in the atmosphere, adding to the total atmospheric CO2 level by c. 66 ppm. For a shift of c. 500 Gt C, the increase of CO2 during the glacial period would be c. 35 ppm, and for around 1700 Gt C it would be c. 115 ppm. If this were the only factor in the carbon cycle to vary between glacial and interglacial conditions, the lower glacial-age land carbon storage would give an atmospheric CO2 level 44-115 ppm higher than during the interglacial. In fact, the CO2 level was some 80 ppm lower during the LGM than during the Holocene. The reduction of CO2 was probably caused by the oceans taking up more carbon under glacial conditions than during interglacial conditions. In general, previous attempts to understand the global carbon cycle during the LGM have only recognized that there was a reduction of 80 ppm atmospheric CO2 under interglacial conditions, requiring the removal of about 170 Gt C from the atmosphere into the ocean. Various scenarios of change in oceanic processes, such as an increase in the 'biological pump' or a slowing of the rate of deep ocean circulation, would allow this extra amount of carbon to be held in the oceans under LGM conditions (reviewed by Broecker & Peng 1993). Unfortunately, if the 'extra' carbon that would enter the land system during the glacial-interglacial transition is added, then there are problems with the ocean models. Adding a modest 60 GtC of land carbon (given a very low estimate of the LGM-Holocene land carbon increase of 425 GtC, and after allowing for the alkalinity buffer), Broecker & Peng (1993) found that several ocean carbon uptake mechanisms working in parallel might have been able to remove this extra carbon. However if in a more realistic quantity such as 140 G t C or more is added (allowing for an increase in land carbon by 1000 GtC), existing models are unable to explain the sink for this carbon during the LGM. It was probably held within the oceans, but existing models do not adequately explain how it was held there. Thus, consideration of the land carbon storage budget illuminates the need for a better understanding of ocean carbon cycle processes during the last glacial. Another result from the consideration of the varying size of the land carbon reservoir is its role in opposing the reduction of CO2 by the oceans under LGM conditions. If the release of CO2 from the land had not occurred during the onset of glacial conditions it can be surmised that the uptake of atmospheric carbon by the oceans would have occurred nonetheless. Without the land source to replenish it, atmospheric CO2 would have been at least several tens of ppm lower than it actually was. Given that the 80ppm lower CO2 level is generally seen as crucial to the
34
J. M. ADAMS & H. FAURE
magnitude of cooling that occurred during the last glacial, an even lower level of C O 2 would have resulted in a greater extent of glaciation. Indeed, since the Earth appears to be continually on the periphery of an ice-up of its surface (Crowley & North 1991; Lovelock & Kump 1994; Kagan 1995), one might regard the extra carbon released by the land during glaciations as preventing a catastrophe caused by excessive ocean carbon uptake.
Surface weathering There is a continual flux of carbon dioxide out of the atmosphere through chemical reactions involving minerals in rocks and soil. The two main types of weathering reaction which take up carbon dioxide are: Silicate weathering, involving reactions such as the following CaSiCO3 + 2CO2 + 2H20 =~ Si(OH)4 + Ca(HCO3)2 and carbonate weathering CaCO3 + CO2 + H20 =~ Ca 2+ + 2HCO 3 The first reaction provides an irreversible sink for carbon dioxide on the glacialinterglacial timescale, whereas the second is much more easily reversible. Nevertheless, both reactions provide a considerable flux of carbon out of the atmosphere each year; only part of which is balanced by dissolution of bicarbonate back into carbonate, carbon dioxide and water. Probst (1992) has calculated that the annual flux of inorganic carbon into rivers from these weathering reactions is 0.39 Gt C/a -1 Although part of this carbon is 'ancient', derived from the carbonate in limestones, for example, most of this is derived more recently from atmospheric CO2 and would be enough to empty the atmosphere of carbon in only a few thousand years if it were not compensated by metamorphic and volcanic output. If the weathering sink varies on a glacial-interglacial timescale this too could have a significant effect on the whole global carbon cycle. Present-day weathering rates are strongly dependent on moisture regime, temperature and rock type, and vegetation and the soil biota catalyse the reactions (Velbel 1993; Ludwig et al. 1995). Given that conditions were generally drier and less vegetated during glacial periods, weathering rates would probably have been lower than during the moist, warm interglacial optima (Adams 1995a). Weathering rate also shows a strong temperature relationship if rock type and water availability are considered (Velbel 1993; Volk 1993), and the globally cooler conditions of the Isotope Stage 2 period (and the previous cool stages 5e-3 together spanning the last c. 100000 years) would have suppressed weathering even further. On the basis of the weathered carbon export in rivers estimated by Probst (1992), Adams (1995b) has estimated the LGM global weathering rate, based on past distributions of biomes according to the QEN Atlas, and suggests that in contrast to the 0.39GtC/a -1 flux at present, LGM weathering uptake would have been about 0.15 Gt C/a -1 (Tables 1 & 2). Although the riverine carbon export rate overestimates total carbon uptake by weathering (and at least some carbon may be released back to the atmosphere by carbonate precipitation on reaching the oceans), the general point holds true. The estimate of the change in area of moist montane environments remains speculative, but a
35
MOISTURE CHANGES & CARBON CYCLE Table 1. Regrouped areas of global vegetation types (• 106 km 2) 18000 years ago 1. 2. 3. 4. 5. 6. 7.
8000 years ago
5000 years ago
0.762 4.235 0.025 8.558 34.672 82.197
7.575 4.905 17.308 16.947 9.947 22.640 21.421
6.701 4.975 17.865 18.273 9.002 26.683 20.036
Mesic taiga Temperate humid Tundra, mid-Taiga Temperate mesic Temperate submesic Sub-desert/dry grassland Absolute desert, lake swamp, ice 8. Tropical seasonal 9. Tropical humid 10. Tropical moist montane
-
24.547 7.502 0.543
24.647 23.229 1.206
23.404 22.615 1.105
Total areas
163.000
150.000
150.000
After Adams (1995b) & QEN data base maps.
difference of this general order remains a realistic scenario because the world's m o u n t a i n ranges and plateaus were generally m u c h drier and colder during the Isotope Stage 2 period and indeed for most o f the 70 000 years preceding this (Li et al. 1995; Crowley & N o r t h 1991). Considering that even a 0.1 G t C / a - l increase in Holocene weathering c o n s u m p t i o n (relative to the Last Glacial) would add up to 1050 G t C over the timespan of about 10 500 calendar years since the early Holocene, this factor has the potential to create some m a j o r 'missing sinks' and 'missing
Table 2. Calculated globalweathering rates based on river bicarbonate transport
1. 2. 3. 4. 5. 6. 7. 8. 9. 10.
Mesic taiga Temperate humid* Tundra, Mid-taiga Temperate mesic Temperate submesic Sub-desert/dry grasslandt Absolute desert, lake swamp, ice Tropical seasonal Tropical humid Tropical moist montane
HCO3km-2a -1 (t)
18000 years ago
12.7 72.3 12.4 31 24.4 2.5
-
0 5.35 16.5 69.9
Total HCO3 export in rivers
55.092 52.512 0.775 208.815 86.672 0
5000 years ago
96.2 354.000 214.61 525.357 242.706 66.500
72.402 359.692 221.526 566.463 219.648 76.707
0
0
131.326 123.783 37.955
131.861 379.500 84.299
125.211 370.147 77.239
696.930
2095.033
2089.032
x 0.196 (mole fraction of C) x 0.001 to convert into Gt Inorganic C export in rivers 0.135 Gt * Includes temperate sub-humid. t Includes both temperate and tropical. After Adams 1995(b) & Probst 1992.
8000 years ago
0.410 Gt
0.409 Gt
36
J. M. ADAMS & H. FAURE
sources' in the global carbon cycle. If metamorphic/volcanic input remained relatively constant throughout this period, then the source of the 'extra' hundreds, or even thousands, of gigatonnes of carbon to sustain atmospheric CO2 levels during interglacials, or the carbon sink during cold, dry glacial periods instead of being taken up in weathering and accumulating as CO2 in the atmosphere, should be identified. It is possible that the late-interglacial pattern of decline in CO2 observed in ice-core data from the Eemian is related to 'exhaustion' of the supply of CO2 by a sustained bout of weathering under interglacial conditions. The problem of balancing the carbon budget on the glacial-interglacial timescale is made even more difficult because when a glacial phase ends, the land surface is 'charged' with weatherable minerals that are unstable under the new, moister and warmer climate regime (Adams 1995a). For example, these minerals would have been in the form of finely divided silicate and carbonate minerals in glacial outwash deposits, desert surfaces and loess deposits, and also soil carbonate in the formerly semi-arid regions that occupied much of the world during the last glacial (Fairchild et al. 1995). Although difficult to quantify, the global uptake of CO2 by weathering during the early stages of moist interglacial conditions could have been much higher than present-day observations on soils would suggest. Furthermore, the reduction of weathering CO2 uptake during glacial conditions would allow the atmospheric CO2 level to remain higher than would otherwise be the case, thereby providing an additional factor which would raise LGM CO2 levels in opposition to oceanic uptake of carbon.
Lake organic sediments Large areas of the Earth's surface are at present covered by freshwater lakes, particularly in the boreal zones of the Northern Hemisphere. Most of the millions of small lakes in the high-latitudes of North America and Siberia have formed since the onset of moist Holocene conditions. Highly organic-rich oozes often reach many metres in thickness, generally having accumulated since the beginning of the Holocene. A large number of small oxbow lakes and layers of organic-rich floodplain sediments also exist in tropical river valleys in the rainforest zones; again these seem to be generally a product of the moist Holocene conditions that replaced the drier Glacial phase. Freshwater lakes did exist during the LGM-Late Glacial period, but they were less abundant. Large areas of the high latitudes were dominated by arid conditions and aeolian erosion (Spasskaya 1992; Velichko & Spasskaya 1992). Notable exceptions were the large proglacial lakes contained by the Laurentide, Fennoscandinavian and Uralic ice sheets (Goncharov 1989; Velichko & Spasskaya 1992). In contrast with most of their present-day boreal counterparts, these lakes were generally characterized by grey or yellow organic-poor sediments due to the combination of very low water temperatures and aridity in their catchments (Goncharov 1989; Spasskaya 1992; Velichko & Spasskaya 1992). If an average organic carbon content of 30 kg/m -2 is assigned to the areas of lake sediments which accumulated since the start of the present interglacial alone, for a global lake area of 2.5 • 106 km 2 (Olson et al. 1983; Adams 1995b) one obtains a newly formed reservoir of 75GtC. This must be offset against the rate of lake organic sedimentation that occurred during the last glacial period, and those glacial
MOISTURE CHANGES & CARBON CYCLE
37
lakes which have emptied and released their organic carbon back into the general carbon cycle. Even without being able to completely compensate for this it is evident that there would still have been a greater rate of organic lake sedimentation during the Holocene as opposed to the LGM-Late Glacial. This increase represents a drain on the ocean-atmosphere carbon reservoirs, which nevertheless has kept the CO2 level higher during Holocene conditions. Further 'missing reservoirs' during glacial conditions are needed to explain the source of these tens of gigatonnes of carbon.
Sea-level fall The land system did not lose water during glacial conditions, it gained it. The extra water was stored in the ice sheets that covered c. 37 x 10 6 km 2 of the Earth's surface, which held an extra r 40 x 10 6 k m 3 of water (Lorius 1991), and in consequence, the sea level fell by 100-120 m; a 3% reduction in ocean volume relative to the present. The freshwater of which ice sheets are composed contains very little carbon and therefore the ocean must have retained all of the carbon that was previously present in this 3 % of its volume, representing a net gain of c. 1200 Gt C (40 000 x 0.03) by the ocean waters that remained. This represents another, very considerable, 'missing' amount of carbon whose disappearance needs to be accounted for if the carbon cycle is to be fully understood, but it is surprisingly not allowed for in any ocean models that seek to explain the carbon cycle during the last glacial period.
An overview: implications of the land system for the carbon cycle In response to large-scale changes in hydrological and temperature regime, the Holocene land system appears to have 'gained' a vast amount of carbon from some unknown source or sources, possibly the ocean (Fig. 4). The bulk organic carbon storage reservoir in vegetation, soils and peats probably gained at least 1000 Gt between LGM and Holocene conditions. The lake-bed and groundwater carbon reservoirs likewise probably accounted for at least several tens of gigatonnes of the Holocene carbon uptake by the land system. Greatly increased weathering rates during moist Holocene conditions seem likely to have taken in hundreds or perhaps thousands of gigatonnes extra carbon, relative to the cumulative average weathering rate during the last glacial period. It is important to note that these changes in land reservoirs and sinks did not operate strictly simultaneously. For example, the lack of a strong (interglacial level) weathering sink during the last Glacial (Isotope Stage 2) period was cumulative and thus reached its peak importance towards the end of this arid period, whereas the land organic carbon reservoir in vegetation and soils was at its low point at the LGM and was already increasing by the end of the last Glacial. Nevertheless, in general these sinks and reservoirs can be seen as combining to highlight the existence of a major 'missing source' of carbon that was not translated into higher glacial CO2 levels. Furthermore, the decrease in ocean volume that was associated with the growth of ice sheets on land would have pushed a further burden of more than 1000 Gt C into the remaining bulk of the oceans. The closer that the carbon cycle on the glacial-interglacial timescale is investigated, the less it seems that anyone truly understands it. The entry of this
38
J.M. ADAMS & H. FAURE
large amount of carbon into the land system under Holocene conditions must be explicable in terms of one or more the following three hypotheses. (i) The operation of unknown mechanisms within the oceans for storing carbon under glacial conditions, in addition to the extra source from the reduction in ocean volume. This mechanism appears the most plausible of the three possibilities at present.
a) INTERGLACIALWORLD Atmosphere 590 GtC (weathering 0.2-0.3 GtC/yr) Holocene land ecosystems>2,000 GtC
Lake beds >75 GtC
Oceans 40
b) GLACIALWORLD Atmosphere 420 GtC (weathering 0.1 GtC/yr) LGM land ecosystems1000 Gt carbon not taken up due to lower weathering rates on land + carbon not in lake sediments >75 GtC.
Fig. 4. Box diagram showing some of the major changes in the carbon cycle between glacial and interglacial conditions. During the arid glacial stage, a large amount of interglacial carbon is 'missing' and is presumably held either in the oceans or under ice sheets.
MOISTURE CHANGES & CARBON CYCLE
39
(ii) Retention of organic carbon from earlier (e.g. Eemian Interglacial, or interstadial) ecosystem cover underneath ice sheets, which was released under during deglaciation, as suggested by Franzen (1994). However, although buried preglacial carbon does occur in some localities, the lack of organic content in most moraine deposits from the last glacial (e.g. West 1978; Williams et al. 1993) contradict this hypothesis. It would appear that most land ecosystem carbon reenters the broader carbon cycle at the onset of glaciation, rather than being covered by ice or held in permafrost. (iii) An increase in volcanism since the beginning of the Holocene, which would supply extra carbon of which a large part was taken up by the land system. There is some evidence for this in the Mediterranean region (mentioned briefly by Caldiera 1992), but its true importance remains unknown.
Implications for the global climate system The hydrological cycle seems to exert a major influence on glacial-interglacial changes in carbon reservoirs through its impact on a mixture of biologically linked and non-biological processes. It appears that the combined effect of reduced land carbon reservoirs, reduced ground water and lake bed reservoirs and weathering rate, would have allowed the CO2 level at the LGM and during the Last Glacial to remain substantially higher than would have been the case if these continental reservoirs and processes had continued in their previous interglacial mode. Thus it seems that there is a feedback between the global climate and the terrestrial carbon cycle; as climate enters its full glacial mode of cold and aridity, the terrestrial system acts as a damper and releases carbon to prevent climate conditions becoming more extreme. Likewise, when climate is in its full interglacial 'optimum' state, the moist, warm conditions favour carbon uptake into the land system, thereby damping the warming. Viewed in this sense, the changes in the terrestrial carbon cycle that occur between glacial and interglacial conditions act as negative feedbacks (damping loops) to stabilize the Earth's climate and atmospheric composition against the positive feedbacks (amplifying loops) which involve ocean carbon uptake, and also the ice sheet albedo on land and sea ice albedo in the oceans. However, it should be considered that land ecosystems also amplify the broad climate oscillations in their own positive feedback loops involving climate. These include their effects on surface albedo, methane and dust contribution to the atmosphere, although the quantitative significance of these effects remains uncertain. The overall picture that one gains from considering the terrestrial system in its various capacities is that even a single component, such as vegetation cover, can simultaneously exert a damping influence (e.g. by taking up CO2 and promoting rock weathering as conditions become moister and warmer) and an amplifying influence (e.g. by increasing its surface coverage and thereby reducing dust fluxes to the atmosphere as global conditions get moister and warmer) on the glacial-interglacial climate cycle. There is much that is still not understood about the whole system, but what is already evident is that land ecosystem processes, and the terrestrial carbon cycle in particular, must play a major role in producing and/or modifying the observed pattern of glacial-interglacial climate oscillations.
40
J. M. ADAMS & H. FAURE
Further research is needed on the processes by which palaeoclimatic variations may have been damped or amplified by their effects on biotic processes if an improved understanding of earth history is to be reached. It is possible that positive and negative feedbacks between the carbon cycle and global palaeohydrology have been important in the evolution of climate on timescales both shorter and longer than the glacial-interglacial cycle considered here. Such processes may also prove to be important in the near future due to anthropogenic modification of climate and ecosystems. This work has been made possible by the support offered by the British Council/CNRS ALLIANCE programme for Anglo-French scientific collaboration. It is also a part of the INTAS Project 'Dynamics of the Terrestrial Biota', funded by the European Community. The Laboratoire de Geologie du Quaternaire (CNRS) also provided vital support during the long, cold arid phase in which this work was unfunded. The logistical skills of L. Faure are also much appreciated.
References ADAMS, J. M. 1995a. Weathering and glacial cycles. Nature, 373, p. 110. - - 1 9 9 5 b . Influence of terrestrial ecosystems on glacial-interglacial changes in the carbon cycle. PhD thesis, University of Aix-Marseille II. & FAURE, H. Global vegetation maps since the Last Glacial Maximum; a resource for archaeologists to use. Journal of Archaeological Science, in press. - & WOODWARD, F. I. 1992. The Past as a Key to the Future; the use o f palaeoenvironmental understanding to predict the effects of Man on the biosphere. Advances in Ecological Research, 22, 257-314. , FAURE, H., FAURE-DENARD, L., MCGLADE, J. M. & WOODWARD, F. I. 1990. Increases in terrestrial carbon storage from the Last Glacial Maximum to the present. Nature, 348, 711-714. ALLEY, R. B., MEESE, R. B.& SHUMAN, A. J. 1993. A new Greenland ice-core record from GISP2 Nature, 366, 443-445. BARNOLA J. M., RAYNAUD, D., KOROKETVICH, Y. S. & LORIUS C. 1989. Vostok ice core: a 160,000 year record of atmospheric CO2. Nature, 329, 408-414. BOLIN, B., DEGENS, E. T., DUVIGNEAD, P. & KEMPE, S. 1977. The global biogeochemical carbon cycle. In: BOLIN, B., DEGENS, E. T., KEMPE, S. & KETNER, P. (eds) The Global Carbon Cycle. SCOPE 13. Wiley, Chichester, 1-53. BROECKER, W. S. 1995. Cooling the tropics. Nature, 376, 212-213. & PENG, T-H. 1993. What caused the glacial-to-interglacial CO2 change? In: HEINMANN, M. (ed.) The Global Carbon Cycle. NATO ASI Series, 15, 95-115. CALDIERA, K 1992. Mount Etna CO2 may affect climate. Nature, 355, 401-402. CHAPELLAZ, J. M., BARNOLA, J. M., RAYNAUD, D., KOROKEVICH, Y. S. & LORIUS, C. 1990. Ice core record of atmospheric methane over the last 160,000 years. Nature, 345, 127-131. - - , BLUNIER, T., RATNAUD, D., BARNOLA, J. M., SCHWANDER, J. & STAUFFER, B. 1993. Synchronous changes in atmospheric CH4 and Greenland climate between 40 and 8 kyr BP. Nature, 366, 443-445. CLIMAP 1976. The surface of the ice age earth. Science, 191, 1131-1136. CROWLEY, T. J. 1994. Pleistocene temperature changes. Nature, 371, 664. - - 1 9 9 5 a . A new reconstruction of ice age terrestrial carbon changes. Abstracts, 14th INQUA Congress, Berlin, 56. - - 1 9 9 5 b . Ice age terrestrial carbon changes revisited. Global Biogeochemical Cycles, 9, 377-389. -& NORTH, G. R. 1991. Palaeoclimatology. Oxford University Press, Oxford. -
-
-
-
MOISTURE CHANGES & CARBON CYCLE
41
CURREY, W. B., DUPLESSY, J.-C., LABYRIE, L. D. & SHACKLETON,N. J. 1988. Changes in the distribution of deep water delta-laC of deep water CO2 between the last glaciation and the Holocene. Palaeoceanography, 3, 327-337. FAIRCHILD, I. J., BRADBY, L. & SPIRO, B. 1995. Reactive carbonate in glacial systems. In: DEYNOUX, M. et al. (eds) The Earth's Glacial Record. IGCP 260. Cambridge University Press, Cambridge, UK, 176-192. FAURE, H. 1989. Changes in the continental reservoir of carbon. Palaeogeography, Palaeoclimatology, Palaeoecology, 82, 47-52. FRANZEN, L. 1994. Are wetlands the key to the ice-age cycle enigma? Ambio, 23, 300-308. FREIDLINGSTEIN, P., DELIRE, C., MUELLER, J. F. & GERARD, J. C. 1992. The climate induced variation of the continental biosphere: a model simulation of the last glacial maximum. Geophysical Research Letters, 19, 897-900. GONCHAROV, S. P. 1989. Last glaciation in western Siberia and glacial dammed lakes in the middle Ensei river basin. [In Russian]. PhD thesis. Institute of Geography, Academy of Sciences, Moscow. HARMON, M. E. & HUA, C. 1992. Coarse woody debris dynamics in two old-growth ecosystems. Bioscience, 41, 604-610. KAGAN, B. A. 1995. Ocean-atmosphere Interaction and Climate Modelling. Cambridge University Press. KOERNER C. & ARNONE, J. A. 1992. Responses to elevated carbon dioxide in artificial tropical ecosystems. Science, 257, 1672-1674. LE HOUEROU, H. N. & HOSTE, C. 1977. Rangeland production and annual rainfall relations in the Mediterranean Basin and in the African Sahelo-Sudanian zones. Journal of Range Management, 3, 181-189. LI, J. et al. & 18 others. 1995. Uplift of the Qinghai-Xizang (Tibet.) Plateau and Global Change. Lanzhou University Press, China. LORIUS, C. 1991. Glaces de L'Antarctique. Editions Odile Jacob, Paris. , JOUZEL, J., RAYNAUD, D. & LE TREUT, H. 1990. The ice-core record: climate sensitivity and future greenhouse warming. Nature, 347, 139-145. LOVELOCK, J. E. & KUMP, J. R. 1994. Failure of climate regulation in a geophysiological model. Nature, 369, 732-735. LUDWIG W., AMIOTTE-SUCHET, P., MUNHOVEN, G. & PROBST, J-L. 1995. Spatial distribution of the consumption of atmospheric CO2 by continental erosion. In Press. MASLIN, M., ADAMS, J. M., THOMAS, E. & FAURE, H. & HAINES-YOUNG, R. 1995. Estimating the carbon transfer between oceans, atmosphere and the terrestrial biosphere since the Last Glacial Maximum: a review of two contrasting methods of estimation. Terra Nova, 7, 358-366. MCCONNAUGHAY, K. D. M., BERNSTON, K. D. M. & BAZZAZ, F. A. 1993. Plant responses to carbon dioxide. Nature, 361, 24. MOONEY, H. A. & KOCH, G. W. 1994. Impact of rising CO2 concentrations on the terrestrial biosphere. Ambio, 23, 74-76. NEFTEL, A. OESCHGER, H STAFFELBACH,T. & STAUFFER, B. 1988 CO2 record in the Byrd ice core 50,000-5,000 years BP. Nature, 331, 609-611. OLSON, J. S., WATTS, J. A. & ALLINSON, L. J. 1983. Carbon in Live Vegetation in Major World Ecosystems. Environmental Sciences Division Publication No. 1997. Oak Ridge National Laboratory, Tennessee. PENG, C. 1994. Reconstruction du stock de carbone terrestre du passe a partir de donnees polliniques et de modelels biospheriques depuis le dernier maximum glaciaire. PhD thesis. Universite d'Aix-Marseille II. , GUIOT, J. & VAN CAMPO, E. 1994. Reconstruction of past terrestrial carbon storage in the Northern Hemisphere from the Osnabroeck Biosphere Model and palaeodata. Climate Research, in press. POST, W. M., EMANUEL, W. R., ZINKE, P. J. & STANGENBERGER,A. G. 1982. Soil carbon pools and world life zones. Nature, 298, 156-159. PRENTICE, I. C., SYKES, M. T., LAUTENSCHLAGER,M., HARRISON, S. P., DENISSENKO, O. & BARTLEIN, P. 1993. Modelling global vegetation patterns and terrestrial carbon storage at the last glacial maximum. Global Ecology & Biogeography Letters, 3, 67-76.
42
J. M. ADAMS & H. FAURE
PROBST, J.-L. 1992. Geochemie et Hydrologie de L'Erosion Continentale. Sciences Geologiques. Memoires 94 (Doctoral thesis), University of Strasbourg. QEN 1995. The Quaternary Environments Network Atlas and Review of Palaeovegetation during the Last 20,000 years. World Wide Web Address; http://www.soton.ac.uk/,-~tjms/ adamsl .html. SHACKLETON, N. J. 1977. C-13 in Uvigerina : Tropical rainforest history and the Pacific carbonate dissolution cycles. In: MALAHOFF, A. & ANDERSON, N. R. (eds) The Fate of Fossil Fuel CO2 in the Oceans. Plenum Press, New York, 401-427. SHELL, D. & PHILIPS, O. 1995. Evaluating turnover in tropical forests. Science, 268, 894. SIEGENTHALER, U. 1989. Glacial-interglacial atmospheric CO2 variations. In: BRADLEY, R. S. (ed.) Global Changes of the Past. UCAR/IES. Boulder, Colorado, 245-278. SPASSKAYA, I. I. 1992. Dominant geomorphic processes during the maximum cooling of the last glaciation. In: FRENZEL, B., PECSI, B & VELICHKO, A. A. (eds) Atlas of Palaeoclimates & Palaeoenvironments of the Northern Hemisphere. INQUA/Hungarian Academy of Sciences. Budapest, 34-35. STARKEL, L. 1989. Global paleohydrology. Quaternary International, 2, 25-33. STREET-PERROTT, F. A. & ROBERTS, N. 1983. Fluctuations in closed-basin lakes as an indicator of past atmospheric circulation patterns. In: STREET-PERROTT, A. et al. (eds) Variations in the Global Water Budget. Reidel, Dordrecht, 331-345. VAN CAMPO, E., GUIOT, J. & PENG, C. 1993. A data-based re-appraisal of the terrestrial carbon budget at the last glacial maximum. Global and Planetary Change, 8, 189-201. VELBEL, M. m. 1993. Temperature dependence of silicate weathering in nature: How strong a negative feedback on long-term accumulation of atmospheric CO2 and global greenhouse warming? Geology, 21, 1050-1062. VELItgHKO,A. A. & SPASSKAYA,I. I. (eds) 1991. Legend of regions of the land during maximum stage of late Valdai glaciation. Geogeodezia SSSR. VOLK, T. 1993. Cooling in the late Cenozoic. Nature, 361, 123. WEBB, R. et al. 1995. Bibliography and inventory of sites with information for constructing digital global maps of vegetation distribution 18,000 yr. BP. NOAA Palaeoclimatological. Publications Series Reports of National Geophysics Data Center, Boulder Colorado, in press. WEST, R. 1978. Quaternary Geology and Biology'. Cambridge University Press, Cambridge, UK. WILLIAMS, D. L., DUNKERLEY, D. L., DE DEKKER, P., KERSHAW, A. P. & STOKES, T. 1993. Quaternary Environments. Edward Arnold, London. WOODWARD, F. I. 1987. Climate and Plant Distribution. Cambridge University Press, Cambridge. WRIGHT, H. E. (ed.) 1993. Global Climates Since the Last Glacial Maximum. University of Minnesota Press, Minneapolis. ZINKE, P. J., STANGENBURGER,A. G., POST, W. M., EMMANUEL,W. R. & OLSON, J. S. 1984. Worldwide organic soil carbon and nitrogen data. Environmental Sciences Division, Publications 2212. Oak Ridge National lab/US Department of Energy.
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes: the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 43-56
Erosion and sediment yield in a changing environment D. E. W A L L I N G
Department of Geography, University of Exeter, Exeter, EX4 4R J, UK Abstract: Existing assessments of patterns of global sediment flux from the land to
the oceans and of global patterns of sediment yield have tended to treat the global denudational system as a static system. There is, however, increasing evidence that the sediment loads of the world's rivers have changed significantly as a result of human activity and interference. As well as being of scientific interest, such changes have important environmental and economic implications and should be included in current concerns for the impact of global change on the Earth's environment. Lack of reliable long-term records precludes a detailed assessment of longer-term and recent changes in suspended sediment transport by world rivers, but the available information can be supplemented by evidence from several other sources, which include the long-term geological perspective, lake sediment records, catchment experiments and space-time substitution; all of which provide valuable information concerning the sensitivity of river sediment loads to environmental change and the magnitude of the changes involved. There is evidence of significant increases and decreases in sediment yields in many areas of the world. Any attempt to relate such changes in sediment loads to environmental change within the upstream drainage basin must, however, take account of the complexity of the sediment delivery system. With the current concern for global change and the impact of both climate change and human activity on the global system, there is clearly a need to consider erosion and sediment yield as a key component of the system and to assess their sensitivity to environmental change. Knowledge concerning changing sediment fluxes is important to studies of rates of landform development, terrestrial inputs to the oceans and global element budgets, but such changes can also have important environmental, economic and social implications relating to land degradation and reduced crop productivity, increasing rates of reservoir sedimentation, destruction of aquatic Table 1. A comparison of soil erosion rates under natural undisturbed conditions and under cultivation in selected areas of the worm
Country
Natural (kg m -2 a-l)
Cultivated (kg m -2 a-l)
China USA Ivory Coast Nigeria India Belgium UK
100 #m) of the sediment. Age determination using luminescence dating is based on the following equation: ED (Gy) age = dose rate (Gy/ka)
ED determ&ation
ED (equivalent dose) is determined using a partial bleach methodology (Wintle & Huntley 1980). The ED is a measure of the dose absorbed by the sediment and this is built up by decay of the naturally occurring radioactive elements 238U, 232Th (and their daughter products) and 4~ which liberates electrons that become trapped in crystal lattice defects within the sediment grain. The number of trapped electrons builds up with time. Stimulation with light (here IR to produce IRSL) or heat (to produce TL) dislodges trapped electrons which recombine at luminescence centres (also crystal defects) producing a photon of light per recombination. The number of photons emitted is therefore a measure of the number of electrons trapped and thus a measure of the dose built up in the sediment. The use of a partial bleach methodology avoids any potential overbleaching (bleaching of luminescence signal beyond that at deposition) which may result if the total bleach or additive dose methodology was applied (cf. fine-grained experiments of Fuller et al. 1994). The light spectrum used in the partial bleaching of this Spanish alluvium was restricted. Berger & Luternauer (1987) indicate the solar spectrum to be severely attenuated below 500 nm and above 690 nm in turbid water akin to that in a river. To avoid overbleaching in the laboratory by using wavelengths not necessarily experienced during transport or at deposition, Berger (1988, 1990) suggests that only wavelengths >550 nm should be used for optical bleaching. A 500-690 nm window is used in this study to construct partial bleach curves (Fig. 2). Bleaching wavelengths were restricted using a Schott BG39 and Corning 3-67 filter in combination. The power of bleaching beneath a solar simulator (SOL2) was further reduced by adding a Kodak-Wratten 0.9ND neutral density filter, achieving a bleaching power of 0.35 mW cm -2 in a SOL2. IRSL measurements were made using a Daybreak 2000manual IRSL system, detecting signal transmitted through a Schott BG39 filter only (50% light transmission between 340 and 600nm); a neutral density filter (0.15 ND) was used in signal 9detection for samples M8, M13, M14, M15 and C1 in which irradiated IRSL signals exceeded 300 000 counts/second (Daybreak photomultiplier tube detection limit). All sample aliquots underwent natural normalization (1.0 s IR stimulation) to remove the effects of inter-disc variation in natural IRSL signal. A 1.0 s IR stimulation using the Daybreak 2000 system causes a less than 1% reduction in signal. Irradiation doses applied were dependent upon sample sensitivity to doses given in preliminary tests;
2soooo
I 100
.[
IIO-
200000 R. 40c o
20-
150000
I
o
m
100000
B 9 o 9 [] []
r,/,j m
50000
2ooo
4000 low iooo BLEACHl"]US(lll~ONI
unbleached 11100 t blench 2300 o blench 3100 m bleach 5000 a bleach 10300 e bleach
'
-150
-100
-50
I
'
i
0
'
50 dose
C2
I
'
100
I
'
I
200
150
(Gy)
100000 SS
2s 2o -
BOO00
W
-
a
-
_ c
o
60000
o m
o: u
0
9 o o o o
40000
.-I r n-
20000
unbleached 1800 e blotch 2300 s blotch 3100 s bleach 5000 s blotch 10300 e bleach
' -40
C7
0000 OOO9 1OO0O ;E (SECONDS)
L~O0 4000 BLEACH
-20
I 0
'
I
'
20
I 40
dose
'
I 60
'
I
80
'
100
(Gy)
Fig. 2. ED determinations: examples of partial bleach methodology are given (C2 and C7). ED variation with bleach time (where appropriate) is shown in the inset graph. Note: these graphs are illustrative, displaying simple curves fitted to mean values. Actual ED calculation used Griin's software (pers. comm. 1991) in which the growth curve is a best fit line accounting for scatter between values.
QUATERNARY
FLUVIAL SEQUENCES, GUADALOPE,
105
SPAIN
maximum beta doses, delivered using a Daybreak irradiator at a 9~176 source strength of 3.8 Gy/minute for coarse grains (> 100#m), varied from 30 to 800Gy. Two representative sample types (C2 and C7) give an indication of the dose regimes applied (Fig. 2). A preheat of 140~ of 62 hour duration was used, removing thermally unstable signal induced following irradiation. Each sample was then stored for 24 hours before measurement (1.0 s IR stimulation) - partial bleach cycles using nine partial bleach times (180, 500, 900, 1300, 1800, 2300, 3100, 5300 and 10 300 seconds duration) reducing IRSL at c. 10% intervals until c. 90% of original (subsequent to preheat) IRSL was removed. The five longest bleaches (_>50% IRSL reduction) were considered in ED calculation (Fig. 2); the EDs determined using shorter bleaches do not give consistent results (Fuller et al. 1994). In most cases, the EDs for each sample fall within errors of each other and a mean value is calculated (cf. Table 1). However, for sample C2, the additive dose ED (intersection with x-axis) is appreciably higher than that of the first three partial bleach EDs shown (Fig. 2). This may be used as a maximum ED, but suggests ED overestimation which the partial bleach methodology seeks to avoid; consequently, in sample C2 and other samples displaying this ED distribution, the first three EDs are used to determine the mean. An additive dose methodological approach was adopted for older samples with exponential growth curves, given the high degree of error in defining the intersect of two exponential curves. A distinction is made between those samples displaying linear growth and those displaying exponential growth (Table 1). Table 1. Additive dose and partial bleach ED values (Gy) for samples collected from the Rio Guadalope Bleaching 0 (AD) time (seconds)
1800
2300
3100
5300
10300
Mas de Las Matas study reach M21 M41 M51 M61 M7 j M8 ~ M91 MI 11 M121 MI3 ~ M14 e M15 e
1.97 4- 0.28* 1.29 4- 0.37* 1.364-0.31" 6.154-0.39" 8.26 4- 0.30 6.10 4- 0.45* 11.31 4-1.11 8.10-1-1.47" 8.34 4-1.49* 75.43 -t- 1.26 65.12 + 2.25* 68.454-21.16" 54.47 4-1.60" 51.89 + 2.37* 51.004-2.10" 147.25 4-18.68" 21.78 4-1.74" 20.31 4- 3.03* 21.43 4- 2.93* 69.70 4-1.99" 21.99 4- 0.50* 21.04 + 1.57" 19.33-t- 1.52" 277.45 4-4.14" 462.844- 107.52" 338.804-91.04 349.444- 90.49 293.50 4-16.39" 229.23 4- 29.81 249.194-44.98
1.37-t-0.36" 6.514-0.42"
1.51 4-0.35* 6.974-0.41
70.77-t-5.19" 51.384-2.17"
71.18-t-1.66 51.544-2.08"
7.354-0.41 10.06 4-1.25 73.31-t-1.51 53.264-2.04*
20.34 4- 2.25* 20.15+ 1.83"
20.82+ 1.20"
21.03-t- 1.11"
374.90 4- 96.50 256.52-t-45.98
410.07 + 100.59 275.60+34.48
444.324- 107.29 286.174-29.30
289.85-t- 13.84 87.75+3.39 53.26 4- 3.03* 49.82 -t- 5.14" 27.224-1.23 28.19 4-1.43" 33.37 4-1.56"
287.36-4-10.47 92.004-2.77 53.56 4- 2.47*
Castelser6s study reach C1 e C21 C41 C51 C61 C71 C81 C91
303.624-8.63* 94.72:t:1.78 54.54 4- 2.32* 49.29 4- 3.31" 28.15 +0.92 28.93 4-1.07" 34.52 :t: 0.94* 67.88 i 1.60"
267.71 + 16.37 266.244- 15.18 260.50-/- 15.16 79.15-t-3.85" 78.364-3.75* 83.00+4.12" 49.23 4- 2.82* 52.74 4- 3.79* 52.19 4- 3.37* 43.79 + 3.90* 24.90+ 1.10" 25.744- 1.17" 26.014-1.11" 26.95 + 1.91" 28.02 5: 1.61" 27.97 4-1.47" 33.49 -t- 1.46" 33.33 4- 1.42 33.74 4-1.53"
1 Denotes linear growth curve; e denotes exponential growth curve. * EDs used (where multiple, value in Table 2 represents mean of those denoted). AD indicates ED obtained by additive dose method.
25.95 + 1.89 29.53 4-1.32" 33.94 -t- 1.01"
106
i.c. FULLER ET AL.
Dosimetry The total dose rate is a function of the sample's radioactivity and it is determined by measuring the constituent external a,/3 and "7 dose rates and internal/3 dose rate. U and Th (and their daughters) emit a, and/3 particles and 3' rays accompanying decay; 4~ emits/3 particles and "7 rays accompanying decay. Measurement of the a activity of the sample in a thick source alpha counter (TSAC) provides a measure of the U and Th contents within the sediment, from which the external a dose rate can be calculated. Beta and ,), dose rates are measured directly in the laboratory and field, using a thick source beta counter (TSBC) (Sanderson 1988) and 7 spectrometer respectively. Internal dose rates (radioactivity from within the grain) are measured by determining K content. Cosmic ray dose rates are estimated using calculations from Prescott & Hutton (1988). Water content (important in the attenuation of radioactivity in the sediment) is estimated based on water content at the time of sample collection (close to zero) and at saturation (measured in the laboratory). Grain size is also considered in the dose rate calculations, being influential in radiation attenuation. The alpha efficiency value is taken as that used in coarse grain feldspar dating (0.2, cf. Mejdahl 1987). A summary of dose rate data is given in Table 2.
I R S L age correction For samples giving luminescence ages over 100 ka, it has been suggested that the age needs to be corrected to take account of long term loss of luminescence signal (Mejdahl 1988; Duller 1994). In this study, correction is applied using the formula: T = r(1 - e -1/~) where T is luminescence age derived from the ED and dose rate, r is the mean life (i.e. lifetime of decay), and t is the geological age. For the current study r is assumed to be 1 million years, as implied for feldspars from Australia (Huntley et al. 1993). No correction for the effect of shallow traps (as reported by Mejdahl & Christiansen 1994) was applied. For the four oldest samples, the corrected ages are given in the final column of Table 3.
Alluvial morphologies M a s de L a s M a t a s A total of 20 terrace surfaces has been identified in the Mas de Las Matas study reach. Six of these relate to the Rio Bergantes, and 14 have been identified in the Rio Guadalope. Heights above river channel vary from 43 m (MTI) to 3 m (MT14) with MT1 standing c. 15 m above the next highest surface, MT2. Correlations between the Guadalope (MT) and Bergantes (B) terraces may be made on the basis of height above modern river level, allowing for variation in gradients between the two systems. It appears that terrace B6 correlates with MT13, B5 with MT10, B4 with MT9, B3 with MT8 or MT7, B2 with MT6 and B1 with MT5. Terraces MT13 (5m), MT9 (7.5m) and MT6 (15m) are the most extensive surfaces and can be traced throughout much of the reach with both the high and low
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
~D
~D
107
I. C. FULLER ET AL.
108
Table 3. Summary of estimated I R S L ages
Sample Terrace unit
Mas de M2 M4 M5 M6 M7 M8 M9* M11 M12 M13 M14 M15
Las Matas MT14 B6/MT13 B6/MT13 B5/MT10 B4/MT9 MT5 MT7 MT7 MTll MT2 MT1 MT2
Castelserdts C1 CT6/fan C2 CT7 C4 CT9 C5 CTll C6" CT8 C7" CT8 C8" CT6 C9 CT7
Grain size (#m)
ED (Gy)
Dose rate (#Gy a-1)
IRSL age (a)
Corrected IRSL age (ka)
125-212 180-212 180-212 125-212 125-150 180-212 125-212 125-212 180-212 180-212 180-212 150-212
1.34+0.35 6.25+0.42 8.22+ 1 . 4 8 68.11+3.2 52.26 + 2.06 147.25• 20.97 + 2.49 69.70 + 1.99 20.73+ 1 . 2 9 277.45 + 44.14 462.84+ 107.52 293.50+ 16.39
3305+313 1882-t-113 2188-t- 146 2526+262 2895 + 380 2717+217 2849 + 272 2844 + 264 2219-t- 167 1505+ 98 2069+ 137 2188 4- 181
405+ 112 3320+229 3756+721 26964-1-3071 18 052 + 2 749 54195+8121 7 360 4- 1 121 24 508 -4-2 380 93424-940 184 352 i 31 698 223702+54034 134141 :i: 13 389
204_38+4~ 2 ~a+72 ""-67 1/i,/+16 *~-15
180-212 180-212 180-212 125-150 180-212 150-212 125-150 180-212
303.62 + 8.63 80.17+3.91 52.59-t-2.97 47.63+ 12.35 25.55+ 1 . 1 3 28.27 4- 1.47 33.73 + 1.32 67.88+ 1 . 6 0
1897 + 137 1000+ 103 1991 + 163 2452+262 1706+ 111 2120 4- 187 2379 + 246 1589• 115
160054 :k 12423 53446+4501 26414-t-2456 19425-t-5447 14976:k 1 178 13 335 + 1 365 14178 :t: 1 567 42718+3253
1,-/A+15 . . . . 14 -
* Tributary fan material overlying terrace surface.
terrace surfaces being fragmentary. Small fans have developed at the m o u t h o f t r i b u t a r y valleys, u p b u i l d i n g existing terrace surfaces and the highest terrace surfaces (MT1 and MT2) m a y well consist partly o f fan material, situated, as they are, close to a large west-bank tributary. A large N W - S E - t r e n d i n g p a l a e o c h a n n e l is preserved in the surface o f MT5; its position and size suggests it is a p r o t o - B e r g a n t e s indicating that the G u a d a l o p e - B e r g a n t e s confluence has m o v e d c. 3 k m d o w n s t r e a m since M T 5 times. T r i b u t a r y valleys in the Mas de Las M a t a s reach generally grade to the m o d e r n channel. This contrasts with the G u a d a l o p e at the Castelser~s study reach, where t r u n k stream incision has outstripped t h a t o f the tributaries, leaving m a n y h a n g i n g (Macklin & Passmore 1995).
Castelser6s Eleven terrace surfaces have been previously identified in the Castelser/ts reach by M a c k l i n & Passmore (1995), o f which six (at 4 9, 12, 17.5, 19 and 23 m above present
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
109
river level) can be traced throughout the reach. Five higher terraces (29, 34.5, 45, 56 and 81 m above river level) occur only at the northern end of the basin. The most extensive units are CT6 (23m), CT7 (19m) and CT8 (17.5m). Prominent relict alluvial fans are developed at the mouths of a number of tributary streams and grade to terrace units CT6 or CT7. Tributary alluvial fan deposition slowed shortly after CT8 formation and as the result cutting of barrancas followed. Incision of the Guadalope has been more rapid than its tributaries, many tributaries have been left hanging above the trunk river level. The lower terraces CT9 (12 m) and CT10 (9 m) are less horizontally extensive than the earlier fills and lie (partly or entirely) within a bedrock trench 150-500 m wide.
Alluvial sedimentology Alluvial deposits at Castelserhs (described by Macklin and Passmore, 1995) and Mas de Las Matas, on the basis of their architecture and sedimentology fall into one of four major lithofacies types. (1) Terrace units characterized by numerous multi-storey lenticular or tabular intersecting sheets of clast supported, matrix rich gravels with minor sand and silt belts and channel fills. Such sediment packages are typical of aggrading, low sinuosity braided river environments (Ori 1982; Miall 1985) and is the most common architectural type in terrace units at both Mas de Las Matas and Castelser/ts. (2) Terrace units characterized by horizontally extensive inclined heterolithic stratification (IHS) (Thomas et al. 1987) and a thick associated fine member. Such features are well developed in CT8 and are consistent with meander lobe development (Campbell & Hendry 1987). In CT9, M T l l and MT14, a lower gravel member is not exposed, but thick upper fine members suggest vertical accretion overbank in a floodplain environment. (3) Terrace units where the lower part of the sequence is dominated by IHS, more typical of a high sinuosity river system; while in the upper part cut and fill multichannel system features predominate. (4) Terrace units characterized by horizontally continuous, relatively thin (< 1 m), flat-bedded relatively angular gravel and minor sand sheets. This type of sequence is often found in a tributary-trunk stream location and represents alluvial fan sedimentation. A notable difference occurs between the sedimentology of higher terrace units and the lower, more recent alluvial fills (MT11 to MT14 at Mas de Las Matas and CT9 to CT11 at Castelser~s). The former are dominated by gravels, with very little fine material, whilst the latter are characterised by sands, with relatively little gravel. This suggests a change in the calibre of sediment delivered to the river, which may relate to the stripping of catchment slopes, exhausting coarse sediment supply, enhancing stream competence and trunk stream-tributary coupling.
Alluvial chronologies Figure 3a shows a schematic cross section of CT6 and later terraces at Castelserfis and Fig. 3b shows a cross section of the complete sequence of alluvial units in the Guadalope at Mas de Las Matas; IRSL ages refer to those in Table 3. Table 4 summarizes the alluvial stratigraphy of both reaches, correlations between
110
I. C. F U L L E R E T AL.
t
t
t
~ ,t2 t
I
I
I
I
i
~ 0
r,,) I =
o
o
t.-
-6
o ~,
~
I x ~ m " 0 CO
~ 0~1
0 ~
~ ~'-
[;~A~I .I~A.LI ~ A o q e
0 ",-
uO
S~llOttl
0
~
LD ~"
" ~"
0 ~
i~' :',
LO CO
0 03
LD L'Xl
0 O,J
LD ,r-
0
LO
0
o
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
o~ -
o~ o
o ~ ~
~
111
o ~ o ~ o ~ o ~ o ~
~ ~o
~:
~ ' ~O ~ ~ ' ~0 ~'~~ ~0 . ' - , ~ ~0 , - " ~ ~0 ' "" ~
~
.Oi~ ,-0
O o +!
~,
+"§ o-~-
-H ,~-
-H ~
-H oo
-H c~
0
~o~
o
A
A
~ ~.~
A
A
0
~
~ ~.~
~ v
~A
(~10
~
~
~
~
O0
I~
I~
'~0 ~'~
["
['f"
[--
[--
[--
~
[--
[--
~
[--[--
o
~
o~ __-H
e~
,~
o,_ ,~ 4-1
+'
+~
-I-I
-n
-H~
0
0
0
~
~
O0
~ A
A
~t'
:,
0
e!, ~ "
-H -H -H ~ ~. "~.
~
~
^
[--
ll2
I. C. F U L L E R E T AL.
Castelser~,s and Mas de Las Matas terraces are made on the basis of height above river level of the terrace surfaces and IRSL ages (where available) of subadjacent units. The earliest alluvial units are CT1-5 in the Castelser/ts reach near Alcafiiz. The surface of these terraces lie between 29 and 81 m above present river level. No datable material was available from these units. Although it is not possible to produce a precise estimate of age for these highest terraces, comparison with the dated MT1 (43 m) at Mas de Las Matas (probably correlative with CT3 (45m) at + 7 2 ka BP and are of Castelser/ts) indicates these were probably deposited before 2 ~-'-'-67 Mid- and Early Pleistocene age (Fig. 4). Freeman (1975) and Alonso & Garz6n (1994) have mapped units in the Rio Jarama (Tagus Basin) at equivalent heights to those at Castelserfis (85 m, 60-70m and 40m). Terraces lying at 40m above river level have also been identified in the Jficar, Serpis and Turia rivers by Pay/t & Walker (1986). The IRSL age estimate of MT1 at Mas de Las Matas suggests aggradation towards the end of (oxygen isotope) stage 8 although the error limits could shift this anywhere between mid stage 6 and stage 9. If we accept 253 ka, this aggradation may have occurred in the transition between an interglacial period and a globally cold, glacial phase in climate, according to oxygen isotope records (Martinson et al. 1987) and ice core records (e.g. Jouzel et al. 1993). Incision of MT1 (Mas de Las Matas) occurred at some point within stage 7. Aggradation of MT2 may have been initiated at the end of stage 7/early stage 6 +40 +16 (204_38 ka) and continued until at least 144_15 ka, (IRSL age at the top of MT2 (24m)) throughout a glacial period (stage 6). An IRSL age of 17414 +15 ka within a fan unit which grades to CT6 (23 m) also indicates extensive aggradation during a glacial phase (stage 6) in the Castelserfis reach, with tributary valleys delivering large quantities of sediment to the trunk river. CT4 and CT5 at Castelserfis have no height correlatives preserved in Mas de Las Matas, with equivalent units probably having been reworked and eroded explaining the large height difference between MT1 (43 m) and MT2 (24 m). Aggradation of MT2 ceased some time after 144ka (Table 4, Fig. 4). An additional terrace is present at Mas de Las Matas, MT3 (22 m), is bracketed between 144 and 122 ka, deposited in the transitional climate between stage 6 and substage 5e. The terrace unit may have been removed from Castelser/ts through subsequent reworking, or may represent a greater complexity of alluvial response to climate change in a confluence zone. Assuming CT7 and MT4 to be correlatives (19 m and 20 m respectively above river +19 level), then aggradation in both reaches began at 122_12 ka towards the end of the thermal maximum at substage 5e (122ka). Palaeoclimate reconstructions from marine pollen records in the Adriatic Sea (Turon 1984) and mires and lakes in France (Pons et al. 1992) show a marked decline in arboreal pollen at the 5d-e boundary. Early substage 5d is characterized by low mean temperature and high total precipitation. At the Cova Negra Palaeolithic site, Valencia, a period of cooler and much wetter climate is also evident, dated to 117 ka (corrected to 124ka) (Pr6szynska-Bordas et al. 1992). Two further IRSL age estimates obtained from CT7 give 53 4- 5 ka and 43 + 3 ka for samples collected at the middle and at the top of CT7 respectively. However, there is evidence for a greater instability in the alluvial system at Mas de Las Matas. The top of MT5 has an IRSL age of 54 4- 8 ka, indicating that while CT7 was aggrading downstream, MT4 at Mas de Las Matas had incised and
QUATERNARY
FLUVIAL
SEQUENCES,
GUADALOPE,
113
SPAIN
50 MTI 9
;>
i
40 30
-
9 -
M'I MTI4t
/ • l qMT9
2
10
-
MTI3
50
100
150
200
250
0
0 350 ka
300
50 40 30 CT6 -
iCT9 -
20
9
10
Ftlq2TI 1
50
2
3
100
4
150
5
200
6
250
0 350 ka
300
7
%
9 _1 ~ 0 50 100 150 200 250 300 ka Fig. 4. A geochronology of alluviation in the Guadalope basin at (a) Mas de Las Matas and (5) Castelser~s, compared with (c) the marine oxygen isotope stratigraphy (after Martinson et al. 1987). I I I I I
I
I It
l,
I
I
r
I
I
l
I
I
I
I
I
i
I
I
~
I
filled (MT5). MT5 aggradation may relate to stage 4-3 transition, with cutting occurring soon after 54 ka (sample taken from capping material), in early stage 3. MT4 was probably incised sometime between 122 and 54 ka. Trenching of CT7 at Castelserfis, however, can be more tightly constrained, occurring after 43 ka. Subsequent infilling of the valley floor occurred (CTS) to 17.5m, which has a similar height to MT6 (15m) at Mas de Las Matas. Both CT8 and MT6 are at present undated, but are bracketed between units for which there is
114
I. C. F U L L E R E T AL.
age control. MT6 was deposited and incised between 53-25 ka and CT8 between 43-26 ka, both probably in stage 3. CT8 represents a change in fluvial style from that of a braided river to an actively migrating coarse grained wandering river overlying a bedrock strath (Macklin & Passmore 1995). The predominance of lateral erosion suggests the Rio Guadalope is close to either static or dynamic equilibrium (cf. Bull 1991). In the Castelser~is reach, all major tributary fans lie above or are truncated by CT8, suggesting very little coarse sediment was being delivered to the trunk stream at this time by these tributaries. This would suggest reduced catchment yields of coarse sediment and this may account for the difference in fluvial style. This change appears to have occurred after 43 ka. Both units (CT8 and MT6) deposited after this period have thick upper fine members. This suggests a change in fine-sediment delivery rates which may relate to a decline in precipitation during this period which is recorded elsewhere in western Europe in the Les Echets and Grand Pile pollen records (Guiot et al. 1989). An increase in aridification suggested by these proxy records would have reduced stream competence, leading to an alluvial system increasingly dominated by fine sediment. Stage 3 was characterized by frequent, small amplitude climate oscillations (Martinson et al. 1987) and in NW Europe, it may be divided into four phases of warm/wet conditions at 59-62 ka, 49-56 ka and 35-42 ka and 28-31 ka, with a cold and arid phase centred on 45 ka (Baker et al. 1993). It is uncertain whether these can apply to southern Europe, given the strong climatic gradient suggested by Rousseau and Puissegflr (1990), however, P6rez-Obiol & Juli~ (1994) identified a clear interstadial event 30 to 27 ka BP from Lake Banyoles in northeastern Spain. The upper fine member of CT9 (12m) at Castelser~is has been dated as 28 + 4ka by Durham Luminescence Research Laboratory (reported in Macklin & Passmore 1995) and in this study to 26 + 2 ka (IRSL age estimates), equating to the end of the interstadial event identified in Lake Banyoles (P~rez-Obiol & Juliet 1994). An IRSL age of 25 + 2 ka has been obtained from a fine member capping MT7 (13 m) gravels, from which artefacts have also been recovered. This corresponds (within error), with the age of CT9. Trenching of MT7 occurred after 24 ka, during stage 2 the next stage of alluviation represented by MT8 is undated, but can be bracketed to sometime between 24 ka and 18 i 3 ka (MT9). It most probably correlates with CT 10, which is bracketed between 26ka and 19ka (CTll). MT9, dated at 18ka, therefore correlates with the lowest terrace mapped in the Guadalope at Castelser~s, CT11, which has an IRSL age of 19 9 5 ka. Stage 2 appears to represent an intensive period of fluvial activity in the Guadalope with at least three phases of alluviation and incision, whereas a single cold stage aggradation has been suggested by previous studies in the region (Pay/t & Walker 1986). The complex nature of climate change during the Last Glacial Maximum is perhaps reflected in the Guadalope record. At Castelser~is a series of fan deposits overlying CT6 and CT8 have IRSL age estimates of 15-13 ka indicating significant tributary delivery of sediment during the Older Dryas stadial. At Mas de Las Matas, alluviation occurred between 18 ka and 9 ka (MT10) (Table 4), but at present no alluvial units of this age have been recognised or dated in the Castelser~s reach. MT11 has been dated to the early Holocene with an IRSL age of 9 • 1 ka. Incision occurred after 9 ka and was followed by a major period of alluviation at Mas de Las Matas (MT13) dated to 3.8 + 0.7 ka (at the base) and 3.3 + 0.2 ka (towards the top),
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
115
with up to 3 m of aggradation in 500 years. In addition there are two minor alluvial fills dated to between 9 and 3.8 ka (MT12) and 405 years (MT14).
Discussion Luminescence dating of alluvial units in the Guadalope basin suggests major aggradation phases in the Mid-Pleistocene occurred during periods of climate transition between interglacials and glacials. This is suggested in terrace units CT7 and MT2, where aggradation began during the climatic deterioration preceding substages 5e and 7a respectively. (Fig. 4). A similar pattern is also evident in MT1, although aggradation is dated to a phase of improving climate between stage 8 and stage 7. The deposition of MT7 and CT9 at 25 and 26 ka, respectively, coincides with a period of climate transition shown by the GRIP oxygen isotope record (Fig. 5). The Lake Banyoles pollen record (Prrez-Obiol and Julifi 1994) shows this was also a wetter phase in northern Spain. Aggradation in the Guadalope basin at c. 19 ka (MT9 and CT11) relates to a short cooling trend at 20 ka, associated with a Heinrich event (H2) in the North Atlantic (Bond et al. 1993). Deposition of MT11 appears to date to climate amelioration in the early Holocene. In the absence of detailed regional land-use or climate records for the later Holocene, at present it is not possible to attribute alluviation at 3.8-3.3 ka and 0.4 ka to climatic or anthropogenic causes. However, introduction of agriculture, destabilizing catchment soils, together with suitable climate conditions providing abundant runoff, will readily supply sediment to the fluvial system prompting aggradation. MT12 may correspond with the Cerezuela unit in the Rio Regallo which is dated (by OSL and radiocarbon dating) to 5840-4780 years aP (Macklin et al. 1994). Between 3 ka and 2.3 ka, 8-9m of aggradation has been found in the Rio Jarama (Alonso & Garz6n 1994). This event is of a similar magnitude, aggrading up to 9 m in 700 years, compared with 3 m in 500 years in the Guadalope. The sedimentology of these comparable events is also similar, consisting of basal gravels overlain by upwardly fining sands and silts, suggesting a similar cause, probably a meandering system. No evidence for equivalent incision is found in the Guadalope, but a correlative incisive period was probably responsible for cutting into MT11 or MT12 in the early Holocene. MT14 is a historic fill. The terrace is limited in extent and the sediment dated represents a flood unit, giving an age of 405 + 112 years (AD 1476-1700). Deposition may be associated with the Little Ice Age climate deterioration. Incision of 2-3 m occurred in the Rio Jarama sometime between 390 + 80 radiocarbon years (AD 1407-1663 cal. (2a)) and the present (Alonso & Garz6n 1994), equivalent (within errors) to incision in the Guadalope following deposition of MT14. Trunk stream alluviation in the Guadalope basin does not appear confined to arid, cold stages, as has been suggested by other workers in Spain investigating less well dated alluvial sequences (e.g. Payfi & Walker 1986), rather it seems to have been initiated by climate deterioration from interglacial or interstadial conditions. A mechanism for this response may be a southward shift in the jet stream, a consequence of which would be an enhanced precipitation delivery to the Mediterranean during the winter (Prentice et al. 1992) and higher rates of catchment erosion and runoff. The proximity of the Guadalope basin to the North Atlantic may be critical in explaining the pattern observed. The GRIP ice core suggests Atlantic temperature to have been highly variable in the Late Pleistocene, linked, primarily, to Heinrich events which
116
I. C. F U L L E R
E T AL.
20 MT5=
=
',
),- r-'-~
MT9 ,, _-
MT11
-5
MTf4~MT13 /
I
I
0
15
-
MT7 i-m-t
I
I
10
I
20
~
30
1
I
I
40
I
I
50
I
I
60
i
I
70
I
I
80
I
I
90
I
I
100
I
I
110
~
0
120 ka 2O
CT7
15 CT9
10 ~,,~ i ', -I
I
0
t
i
I
10
~
CT11
I
20
I
t
30
I
i
40
I
r
50
I
i
I
60
I
I
70
J
80
I
I
90
I
I
100
[
I
110
| I
.~, .5 ~
0
120 ka
8180 stage 1
2
3
5a
4
5b
5c
5d
5e
-33 -35 -37 -39
0
-43 I
0
I
10
I
20
I
t
30
2
1
I
1
40
I
I
50
I
I
60
3
I
I
70 4
I
80 5a
j
5b
I
I
90 5c
|
100 5d
I
I
110 5e
jl
-41 120 ka
0
200 400 600
:--2
800 t
0
10
l
20
i
I
30
i
I
40
t
I
50
i
i
60
70
80
90
100
10
1000 120 k a
Fig. 5. A geochronology of alluviation in the Guadalope basin at (a) Mas de Las Matas and (b) Castelserfis, compared with (e) the oxygen isotope stratigraphy in the high resolution GRIP ice core and (fl) the Lac du Bouchet magnetic susceptibility record (after Thouveny et al. 1994).
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
117
cause a sudden cooling, although subsequent warming is also rapid. The role of thermohaline circulation is thought to be critical in forcing these temperature shifts, whereby heat advection from the tropics is increased in response to a reduced meltwater flux following ice sheet collapse (Paillard & Labeyrie 1994). Therefore, even during general cooling, increased evaporation from the North Atlantic may be sufficient to deliver additional precipitation to the northern parts of the Iberian peninsula. This would be likely to promote aggradation, by causing major erosion of catchment slopes unprotected by vegetation which may be unable to respond to such short-term oceandriven climate pulses. In summary, it is proposed that major valley alluviation in the Guadalope basin tends to be associated with transitional phases of climate, during the last interglacial-glacial cycle and possibly in earlier periods of the Pleistocene. This contrasts with interpretations of Pleistocene alluvial sequences elsewhere in Spain and Europe, notably the Thames and Rhine in which gravel aggradation episodes are apparently confined to cold stages (cf. Brunnacker et al. 1982; Bridgland 1988; Gibbard 1989). However, a lack of detailed chronometric control and the fragmentary nature of the alluvial deposits in these systems may conceal responses similar to those identified in the Guadalope catchment.
Conclusions Geomorphological investigations of Quaternary fluvial deposits in the Guadalope valley, northeast Spain, have revealed a record of alluvial response to long-term environmental change during the Mid- and Late Pleistocene, Holocene and historic periods. Eleven alluvial terraces, ranging from 4 to 81 m above river level, have been identified along a 13 km long reach of the Rio Guadalope south of Alcafiiz. Twenty surfaces have been identified in a 4 km long confluence zone with the Rio Bergantes, ranging from 3 te 43 m above river level. Luminescence dating suggests that deposition of these alluvial units dates back to the Mid-Pleistocene. The earliest fluvial unit is dated to the Mid-Pleistocene (probably stage 8 glacial). A complex sequence of large scale cutting and filling occurs up to stage 5. A major phase of aggradation from substage 5e to early stage 3 can be identified at Castelserfis, whilst at Mas de Las Matas during this period two cut and fill episodes may be identified. Following 43 ka, four episodes of cutting and filling are documented in both reaches during the latter part of stage 3 and into stage 2. Four Holocene alluvial units are evident: the oldest deposited at 9 ka, a second unit dated before 4 ka and a third to 3.8-3.3 ka; a phase of minor alluviation is dated to the sixteenth century. Region-wide aggradation/incision events can be tentatively established by comparing the Guadalope response with nearby catchments (notably the Jarama and Jficar) during both the Late Pleistocene and Holocene. However, the lack of dating control in the latter catchments does not enable detailed correlation between alluvial units to be made with the Guadalope at present. Luminescence dating has built up a detailed geochronology of the alluvial response to climate change in the Rio Guadalope spanning the last 250 000 IRSL years. It has permitted, for the first time, a relatively detailed view of river response to environmental change in previously poorly documented periods in the Iberian Peninsula, notably during the last interglacial-glacial cycle, as well as providing a window into the Mid-Pleistocene.
118
I. C. FULLER ET AL.
We wish to thank Carlos A. Navarro for his outstanding assistance in the field and members of the Aberystwyth Luminescence Research Group for their assistance in application of IRSL and criticism of this research. I.C.F. was supported by NERC studentship GT4/92/23/G. For I.C.F., P.A.B., J.L. and A.G.W., this is publication number 382 of the Institute of Earth Studies, Aberystwyth.
References ALONSO, A. & GARZON, G. 1994. Late Quaternary evolution of a medium sinuosity gravel bed river in central Spain. Terra Nova, 6, 465-475. BAKER, A., SMART, P. L. & FORD, O. C. 1993. Northwest European palaeoclimate as indicated by growth frequency variations of secondary calcite deposits. Palaeogeography, Palaeoclimatology, Palaeoecology, 100, 291-301. BERGER, G. W. 1988. Dating Quaternary events by luminescence. In: EASTERBROOK, D. J. (ed.) Dating Quaternary Sediments. Geological Society of America Special Papers, 227, 13-50. - - 1 9 9 0 . The effectiveness of natural zeroing of the thermoluminescence in sediments. Journal of Geophysical Research, 95, 12375-12397. --& LUTERNAUER,J. J. 1987. Preliminary fieldwork for thermoluminescence dating studies at the Fraser River delta, British Columbia. Geological Survey of Canada Papers, 87-1 A, 901-904. BOND, G., HEINRICH, H., BROECKER,W., LAURENT, L., MCMANUS, J., ANDREWS,J., HUON, S., JANTSCHIK, CLASEN, S., SIMET, C., TEDESCO, K., KLAS, M., BONANI, G. & IvY, S. 1992. Evidence for massive discharges of icebergs into the North Atlantic ocean during the last glacial period. Nature, 360, 245-249. ~, BROECKER, W., JOHNSEN, S., MCMANUS, J., LABEYRIE, L., JOUZEL, J. & BONANI, G. 1993. Correlations between climate records from North Atlantic sediments and Greenland ice. Nature, 365, 143-147. BRIDGLAND, D. R. 1988. The Pleistocene fluvial stratigraphy and palaeogeography of Essex. Proceedings of the Geologists' Association, 99, 315-333. BRUNNACKER, K., LOSCHER, M., TILLMANS, W. & URBAN, B. 1982. Correlation of the Quaternary terrace sequence in the Lower Rhine Valley and Northern Alpine Foothills of Central Europe. Quaternary Research, 18, 152-173. BULL, W. B. 1991. Geomorphic Responses to Climatic Change. Oxford, Oxford University Press. CAMPBELL, J. E. & HENDRY, H. E. 1987. Anatomy of a gravelly meander lobe in the Saskatchewan River, near Nipawin, Canada. In: ETHERIDGE, F. G. FLORES, R. M. & HARVEY, M. D. (eds) Recent Developments in Fluvial Sedimentology. Society of Economic Paleontologists and Mineralogists Special Publications, 39, 179-189. DANSGAARD, W., JOHNSEN, S. J., CLAUSEN, H. B., DAHL-JENSEN, D., GUNDESTRUP, N. S., HAMMER, C. U., HVIDBERG, C. S., STEFFENSEN, J. P., SVEINBJORNSDOTTIR, A. E., JOUZEL, J. & BOND, G. 1993. Evidence for general instability of past climate from a 250-kyr ice-core record. Nature, 364, 218-220. DITLEFSEN, C. 1992. Bleaching of K-feldspars in turbid water suspension: A comparison of photo- and thermoluminescence signals. Quaternary Science Reviews, 11, 33-38. DULLER, G. A. T. 1994. Luminescence dating using feldspars: A test case from Southern North Island, New Zealand. Quaternary Geochronology (Quaternary Science Reviews), 13, 423-427. FREEMAN, L. G. 1975. Acheulian sites and stratigraphy in Iberia and the Maghreb. In:. BUTZER, K. W & ISAAC, G. L. (eds) After the Australopithecines; Stratigraphy, Ecology and Culture Change in the Middle Pleistocene. Aldine, Chicago, 661-743. FULLER, I. C., WINTLE, A. G. & DULLER, G. A. T. 1994. Test of the partial bleach methodology as applied to the IRSL of an alluvial sediment from the Danube. Quaternary Geochronology (Quaternary Science Reviews), 13, 539-543. GIBBARD, P. L. 1989. The geomorphology of a part of the Middle Thames forty years on: a reappraisal of the work of F. Kenneth Hare. Proceedings of the Geologists" Association, 100, 481-503.
QUATERNARY FLUVIAL SEQUENCES, GUADALOPE, SPAIN
119
GUIOT, J., PONS, A., DE BEAULIEU, J. L. & REILLE, M. 1989. A 140,000-year continental climate reconstruction from two European pollen records. Nature, 338, 309-313. HUNTLEY, D. J., HUTTON, J. T. & PRESCOTT, J. R. 1993. Optical dating using inclusions within quartz grains. Geology, 21, 1087-1090. JOUZEL, J., BARKOV, N. I., BARNOLA, J. M., BENDER, M., CHAPPELLAZ, J., GENTHON, C., KOTLYAKOV, V. M., LIPENKOV, V., LORIUS, C., PETIT, J. R., RAYNAUD, D., RAISBECK, G., RITZ, C., SOWERS,T., STIEVENARD,M., YIOU, F. & YIOU, P. 1993. Extending the Vostok ice-core record of palaeoclimate to the penultimate glacial period. Nature 364, 407-412. KEIGWIN, L. O., CURRY, W. B., LEHMAN, S. J. & JOHNSEN, S. 1994. The role of the deep ocean in North Atlantic climate change between 70 and 130kyr ago. Nature, 371, 323-326. MACKLIN, M. G. & PASSMORE,D. G. 1995. Pleistocene environmental change in the Guadalope Basin, Northeast Spain: Fluvial and Archaeological records. In" LEWIN,J. MACKLIN, M. G & WOODWARD, J. C. (eds) Mediterranean Quaternary River Environments. Balkema, Rotterdam, 103-113. - - , STEVENSON,A. C., DAVIS, B. A. & BENAVENTE,J. 1994. Responses of rivers and lakes to Holocene environmental change in the Alcafiiz region, Teruel, northeast Spain. In: MILLINGTON, A. C. & PYE K. (eds) Effects of Environmental Change in Drylands. Wiley, Chichester, 113-130. MARTINSON, D. G., PISIAS, N. G., HAYS, J. D., IMBRIE, J., MOORE, T. C. JR. & SHACKLETON, N. J. 1987. Age dating and the orbital theory of the Ice Ages: Development of a high resolution 0 to 300,000-year chronostratigraphy. Quaternary Research, 27, 1-29. MEJDAHL, V. 1987. Internal radioactivity in quartz and feldspar grains. Ancient TL, 5, 10-17. - - 1 9 8 8 . Long term stability of the thermoluminescence signal in alkali feldspars. Quaternary Science Reviews, 7, 357-360. - - & CHRISTIANSEN,H. H. 1994. Procedures used for luminescence dating of sediments. Quaternary Geochronology (Quaternary Science Reviews), 13, 403-406. MIALL, A. O. 1985. Architectural-element analysis: a new method of facies analysis applied to fluvial deposits. Earth Science Reviews, 22, 261-308. ORI, G. G. 1982. Braided to meandering channel patterns in humid-region alluvial fan deposits, River RenD, Po plain (northern Italy). Sedimentary Geology, 31,231-248. PAILLARD, D. & LABEYRIE, L. 1994. Role of the thermohaline circulation in the abrupt warming after Heinrich events. Nature, 372, 162-164. PAYA, A. C. & WALKER, M. J. 1986. Palaeoclimatological oscillations in continental Upper Pleistocene and Holocene formations in Alicante and Murcia. In: LOPEZ, V. F. (ed.) Quaternary Climate in the Western Mediterranean. Universidad Autbnoma, Madrid, 365-376. PI~REZ-OBIOL, R. & JULIA, R. 1994. Climatic change on the Iberian Peninsula recorded in a 30,000-yr pollen record from Lake Banyoles. Quaternary Research, 41,91-98. PONS, A. & REILLE, M. 1988. The Holocene and Upper Pleistocene pollen record from Padul (Granada, Spain): a new study. Palaeogeography, Palaeoclimatology, Palaeoecology, 66, 243-263. , GUIOT, J. L. & REILLE, M. 1992. Recent contributions to the climatology of the last glacial-interglacial cycle based on French pollen sequences. Quaternary Science Reviews, 11, 439-448. PRENTICE, I. C., GUIOT, J. & HARRISON, S. P. 1992. Mediterranean vegetation, lake levels and palaeoclimate at the Last Glacial Maximum. Nature, 360, 658-660. PRESCOTT, J. R. & HUTTON, J. T. 1988. Cosmic ray and gamma ray dosimetry for TL and ESR. Nuclear Tracks and Radiation Measurements, 14(1]2), 223-227. PROSZYNSKA-BORDAS,H., STANSKA-PROSZYNSKA,W. & PROSZYNSKI, M. 1992. TL dating of river terraces with fossil soils in the Mediterranean Region. Quaternary Science Reviews, 11, 53-60. RENDELL, H. M., CALDERON, T., MILLAN, A., PEREZ-GONZALEZ, A., GALLARDO, J. & TOWNSEND, P. D. 1994. Thermoluminescence and optically stimulated luminescence dating of Spanish dunes Quaternary Geochronology (Quaternary Science Reviews), 13, 429-432.
120
I. C. FULLER ET AL.
ROUSSEAU, D.-D. & PUISSEGI)R, J-J. 1990. A 350,000 year climatic record from the loess sequence of Achenheim, Alsace, France. Boreas, 19, 203-216. SANDERSON, D. C. W. 1988. Thick Source Beta Counting (TSBC): A rapid method for measuring beta dose-rates. Nuclear Tracks and Radiation Measurements, 14, 203-207. THOMAS, R. G., SMITH, D. G., WOOD, J. M., VISIER, J., CALVERLEY-RANGE,E.A & KOSTER, E. H. 1987. Inclined heterolithic stratification- terminology, description, interpretation and significance. Sedimentary Geology, 53, 123-179. THOUVENY, N., DE BEAULIEU, J-L., BONIFAY, E., CREER, K. M., GUIOT, J., ICOLE, M., JOHNSEN, S., JOUZEL, J., REILLE, M., WILLIAMS, T. & WILLIAMSON, D. 1994. Climate variations in Europe over the past 140 kyr deduced from rock magnetism. Nature, 371, 503-506. TURON, J. L. 1984. Direct land/sea correlations in the last interglacial complex. Nature, 309, 673-676. WINTLE, A. G. 1993. Luminescence dating of aeolian sands: an overview. In: PYE, K. (ed.) The Dynamics and Environmental Context of Aeolian Sedimentary Systems. Geological Society, London, Special Publications, 72, 49-58. & HUNTLEY, D. J. 1980. Thermoluminescence dating of ocean sediments. Canadian Journal of Earth Sciences, 17, 348-360. -
-
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 121-137
Magnitude and frequency of Holocene palaeofloods in the southwestern United States: A review and discussion of implications Y E H O U D A E N Z E L 1, LISA L. ELY 2, P. K Y L E HOUSE 3, & VICTOR R. B A K E R 3
1Institute of Earth Sciences and Department of Geography, Hebrew University, Jerusalem, 91904 Israel 2 Department of Geology, Central Washington University, Ellensburg, WA 98926, USA 3Arizona Laboratory for Palaeohydrological and Hydroclimatological Analysis, Department of Geosciences, University of Arizona, Tucson, A Z 85721, USA Abstract: Data about the magnitude and time of occurrence of palaeofloods from the lower Colorado River basin enable us to test two long-standing hypotheses which have affected many studies and applications in the field of flood hydrology. The two hypotheses are (a) the existence of an upper boundary to flood magnitudes and whether there is a possibility of determining it from the existing data, and (b) the random occurrence versus clustering of the large floods through time. Earlier observations on regional flood envelope curves indicated the existence of an upper limit for flood magnitudes, but these studies limited their conclusions because of the short length of the systematic gauged data. This limitation is overcome here because palaeoflood data cover a much longer period of observation. Palaeoflood studies provide information about the largest individual floods experienced in many rivers in a specific region occurring over the last millennia. In the southwestern US, this information demonstrates that, even when the length of observational data increases to centuries and millennia, there is no change in the stabilized, regional envelope curves constructed from gauged and historical flood records. This pattern supports the hypothesis of an upper limit to flood magnitudes and points to a method for testing this hypothesis in other regions. There are surprising similarities between the envelope curve of the palaeoflood data and the envelope curve for the gauged and historical data in the lower Colorado River basin. These similarities indicate that in regions of the world where flood data is sparse envelop curves based on palaeoflood studies can provide basic data for engineering design purposes and other hydrological applications. The random occurrence of large floods in time is tested by constructing chronologies for the largest palaeofloods in several basins in the lower Colorado River basin. These chronologies indicate a clustering of the large floods in specific time periods. The similarity between the various time periods characterized by high- and low-flooding and other palaeoclimatic indicators from the southwestern Untied States seems best explained by a climatic control on flood frequency over the last 5000 years.
Several assumptions have u n d e r p i n n e d scientific thought in the research field o f statistical flood hydrology for m a n y years. A m o n g those which are c o m m o n l y used in hydrology are (a) the existence of an upper b o u n d a r y to flood magnitudes, a n d
122
Y. ENZEL E T A L .
(b) the random (v. clustered) occurrence of large floods through time. The difficulties in resolving which and/or whether these concepts are valid arose from the limited lengths of the flood records. These records are usually short, in most cases shorter than 100 years. Here we overcome this limitation by adding regional palaeoflood data that span the last several centuries to millennia. This addition enables us to detect patterns in both the temporal distribution and the variations in the magnitudes of the largest floods. Agreement or disagreement of the regional flood records with the above assumptions will result in the acceptance or rejection of the assumptions. A limited number of regions in the world have sufficient palaeoflood information to apply this approach. By far the largest available palaeoflood data set from a single region is from the lower Colorado River basin in Arizona and southern Utah, which will be the focus of this review.
Palaeoflood hydrology - methodology and available data Fine-grained flood deposits and other palaeostage indicators provide a long record of large floods (Baker 1987a). During large floods in stable bedrock canyons, finegrained sand and silt fall rapidly out of suspension in areas of markedly reduced flow velocity, such as back-flooded tributaries and eddies at channel irregularities (Kochel & Baker 1982; Kochel et al. 1982; Baker 1987a). At some sites, stratigraphic records of multiple floods span several centuries or millennia, and the individual flood deposits can be distinguished through sedimentological criteria. Several types of information can be extracted from such field settings: (a) the minimum stage of the largest flood, (b) a chronology and a minimum stage estimates of other large floods, and (c) estimated magnitudes of the floods represented in the stratigraphic record. This last type of data can be extracted only if the setting and channel morphology are appropriate for hydraulic modeling, which is used to produce instantaneous peak discharge estimates from the minimum stage estimates derived from the heights of the deposits (Baker 1987a; O'Connor & Webb 1988). In this study we are interested in two aspects of the palaeoflood records. The discharge of the single largest flood in each basin will be compared with the largest floods in the modern instrumental record and used to address the question of the existence of an upper boundary to flood magnitudes in the region. Second, the chronologies of the largest floods in each basin are used to produce a regional data set of palaeofloods and examine the temporal distribution of large floods throughout the region. Information on both aspects from as many rivers as possible in the lower Colorado River basin is required to address these questions effectively. Figure 1 shows sites that are suitable for producing chronologies, sites suitable for indicating magnitudes of the largest floods, and sites that are useful for both applications. Evidence of the largest floods is selectively preserved, deposits from smaller floods lie closer to the active river channel and are more likely to be removed by subsequent erosion (Ely & Baker 1990). The ages of the floods were largely determined by radiocarbon dating of associated organic or archaeological material (e.g., Enzel et al. 1994). These dates establish the chronology of the individual palaeoflood deposits preserved in a site and aid in determining the total length of the record chronicled by the deposits. Palaeoflood discharge estimates are calculated by comparing the heights of flood palaeostage indicators with water-surface profiles calculated by the step-backwater
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
123
Fig. 1. Map showing the location of palaeoflood sites in Arizona and southern Utah, which yielded data (a) only for palaeoflood chronologies (open triangles); (b) only for maximum of peak discharges of palaeofloods (squares); and (e) for both chronologies and maximum peak discharges of palaeofloods (circles). The letters are in reference to Table 1. The inset map shows the entire Colorado River basin.
124
Y. ENZEL ETAL.
Table 1. A summary of the single largest palaeoflood magnitude from several drainage basins in the lower Colorado River (Enzel et al. 1993) River/site*
Code Drainage in area Fig. 1 (km 2)
Maximum peak discharge of palaeoflood (m3s-1)
Length of record (years)
Data source
Colorado River, AZ
C
279 350
13 600-14200
c. 4000
Verde River, AZ
V
14 240
5 000
2000
Salt River, AZ
S
11 150
4100-4 600
2000
Salt River, AZt
P
33 650
Tonto Creek, AZ
T
1 630
8 500-9 900 11 300-12700 800-1 000
O'Connor et al. (1994) Ely & Baker (1985); O'Connor et al. (1986a) Partridge & Baker (1897); O'Connor et al. (1986a) Fuller (1986)
c. 500
Aravaipa Creek, AZ Redfield Creek AZ Oak Creek, AZ
A R O
3 160 285 1 213
970 350-400 1 350
c. 900 c. 1000 >>100
Virgin River East Fork, UT
E
840
800-850
1000+
Virgin River, AZ Kanab Creek, UT
Vi K
10 306 5 370
1 700-1 900 400-600~
1000+ c. 500
Escalante Riverw UT Es Es Es Es Boulder Creek, UT B
820 1 900 3 290 4430 450
700-750 1250-1 550 1 850-2 100 860-940 350-450
1000+
Tortolita Mts, AZ
1000+
500+
TM
Cochie Wild Burro
C1 WB WB Ruelas RU Prospect Pr Cafiada Agua Ca White Tank Mts, AZ WT
Tiger Wash, AZ
TW
Sierra Estrella, AZ
SE
9.8 11.1 18.1 6.0 9.6 4.7 14.6
220
2.8
60-80 120-150 200-300 80-100 40-50 30-50 57-142
283-382
21-29
O'Connor et al. (1986a); Ely et al. (1988) Roberts (1987) Wohl (1989) Melis (1990); Melis (pers. comm.) Enzel, Ely & Webb (unpublished data) Enzel et al. (1994) Smith (1990); Smith (pers. comm) Webb (1985); Webb and Baker (1987); Webb et al. (1988) O'Connor et al. (1986b) House (1991); House (unpublished data) House et al. (1991)
600+ House (unpublished data); CH2MHill & French (1992) House (unpublished data), CH2MHill & French (1992) House (unpublished data), CH2MHill & French (1992)
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
125
Table 1. (conth~ued)
River/site*
Code Drainage in area Fig. 1 (km2)
Maximum peak discharge of palaeoflood (m3s-1)
Martinez-Goytre et al. (1994)
Santan Catalina Mts. SC Cafiada Del oro Sutherland Pima Tanque Verde Youtcy Buehman Edgar Alder
Largest Data source flood (years BP)
80 55 10 110 24 103 74 46
110 95 55 240 50 270 130 50
* See Fig. 1 for locations. The palaeofloods are plotted in Figs 1 2, and 3 according to site code except for the Tortolita Mts. palaeoflood estimates which are represented in Fig. 1 as TM. t Salt River downstream of the confluence of the Verde River; expanding flow can cause large overestimation. The actual largest flood was larger than reported by Smith (1990). He identified in the field evidence for a larger flood located in a channel reach which is difficult to model (Smith, pers. comm. 1991). wDifferent sites on the river.
method (O'Connor & Webb 1988). The elevation of a given deposit provides a minim u m estimate for the peak stage of the associated flood, though the underestimation is not large. In many cases, the heights of the deposits closely approximate the actual stage of the flood peak (Ely & Baker 1985; O'Connor et al. 1986a; Baker 1987a; Partridge & Baker 1987; Kochel & Ritter 1987; Baker & Kochel 1988). Kochel (1980) estimated that deposit height was 10% less than actual water surface elevation and Webb (pers. comm. 1992) and Greenbaum (pers. comm. 1992)have documented silt lines and other high water marks 50 to 90 m higher than associated deposits. In several of the palaeoflood studies cited in Table 1 the results are based on silt lines, scour lines, or debris that indicate maximum flood stage. Thus, the quoted discharges are directly associated with the largest flood that has occurred at the site over the period of the palaeoflood record. While we stress that the palaeoflood discharge studies based solely on the height of flood deposits are minimum estimates of the peak discharge, considerable experience demonstrate that discharge underestimation is probably 20% or less (e.g., Kochel et al. 1982). Baker (1987a) reviews the field observations that justify this conclusion. The palaeoflood methodology used for data reported herein corresponds strictly to the 'slackwater deposit and palaeostage indicator' (SWD-PSI) technique (Baker 1987a) applied either to stable-boundary reaches or to reaches with well-known geometry. Data obtained through other palaeoflood reconstruction techniques, including regime-based procedures and palaeocompetence studies, are not included in Table 1. In that table the discharges are listed according to the ranges reported by the original researchers and they are the largest palaeoflood discharges in each of the studied basins in Utah and Arizona (Fig. 1). All of these palaeoflood sites were
126
Y. ENZEL E T A L .
studied by past and present researchers at the Arizona Laboratory for Paleohydrological and Hydroclimatological Analysis in the Department of Geosciences at the University of Arizona. Therefore, we are sure that the field and laboratory procedures used to extract the data were very similar in all cases. As no evidence was mentioned by the original authors as to causes of floods other than precipitation, we assume that all the listed palaeofloods were formed through rainfall-runoff processes. Thirteen of these palaeofloods are the largest in at least the last 500 years, and eight are the largest during the last 1000-4000 years (Table 1).
Upper boundary to flood magnitudes
The hypothesis The perspective that there should be a limit to flood magnitude and that this upper bound is related to the area of the specific drainage basin can be traced to classical studies in hydrology, including the seminal work by Horton. Horton (1936, pp. 437438) was probably the first to hypothesize about the existence of an upper limit to flood magnitudes related to basin size: Flood magnitudes always continue to increase as the recurrence interval increases, but they increase toward a definite limit and not toward infinity. This is believed to be the more rational from of expression. No terrestrial stream can produce an infinite flood. A small stream cannot produce a major Mississippi River flood, for much the same reason that an ordinary barnyard fowl cannot lay an egg a yard in diameter: it would transcend nature's capabilities under the circumstances. Since then, many studies have shown that floods are related to drainage area and that this variable is important in predicting flood magnitudes (e.g., Dooge 1986). In the southwestern United States, Benson (1964) concluded that drainage area is by far the most important basin characteristic in estimating flood magnitudes. For Arizona, his conclusion was supported by Roeske (1978, p. 33), who showed that drainage area is the only statistically significant variable. However, most researchers have used the drainage basin area as a parameter only in estimating magnitude of rare flood, not the magnitude of the maximum expected flood. On the other hand, envelope curves encompassing all the maximum flood peaks (discharge plotted vs. area) experienced either regionally (e.g., Creager 1939; Crippen & Bue 1977; Georgiadi 1979; Crippen 1982) or globally (Costa 1987) were used to advocate the existence of a maximum flood per drainage area. The existence of an upper limit to flood magnitude is not an assumption restricted to the advocates of the utility of envelope curves, the concept also underlies the more common, deterministic approach taken in using rainfall-runoff models of the 'worst case scenarios' for design purposes. An interesting pattern emerges from the construction of regional envelope curves of maximum discharges. Several researchers have noted that incremental increases in the temporal and spatial base of the observational record impart progressively smaller changes in the form and position of envelope curves of peak discharge vs. drainage area (e.g., Creager 1939; Matthai 1969; Crippen 1982; Wolman & Costa 1984; Costa 1987). While this phenomenon can be explained by probabilistic
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
127
reasoning (Yevjevich & Harmancioglu 1987), it has also been hypothesized that the apparent recent stabilization in the envelope curves encompassing the maximum floods in the Untied States is indicative of the existence of an upper limit to flood magnitudes (Wolman & Costa 1984; Costa 1987) and is not simply a stochastic phenomenon. Wolman & Costa (1984) also hypothesized that hydroclimatological processes and basin characteristics sustain the main control on the upper limit of flooding. The hypothesis of an upper limit to flood magnitudes has been difficult to substantiate because of obvious limitations on the rate of accumulation of observational data. In this study we present a means of overcoming the above limitations. Rather than waiting a very long time for sufficient future data, we accumulated data on floods which have already occurred in the region. Thus results from 25 palaeoflood hydrological studies in the lower Colorado River basin covering the last several centuries to millennia are added to the database previously composed of only modern and historical data covering 100 years at best. This augmentation of the flood record extends the effective length of observation at individual sites by hundreds to thousands of years and thereby allows for an independent evaluation of the hypothesis that an enveloping curve with a sufficiently broad spatial and temporal data base stabilizes at a position approximating a natural upper bound to flood magnitudes in a given region. There are two possible results from this exercise: either the additional palaeoflood data will alter (i.e., raise) the regional envelope curve based on modern and historical data, or the maximum palaeoflood discharges will fall within the existing curve. If the curve remains unchanged by the addition of palaeoflood data, the results then are consistent with the hypothesis of an upper limit to flood magnitudes.
Data and results from the southwestern United States The US Bureau of Reclamation (1990) constructed an envelope curve for the largest gauged and historical floods in the Colorado River basin (Fig. 2 curve C) and we use this as the modern envelope curve. The 26 curve-controlling floods are marked in Fig. 2. This curve was originally constructed only for drainage basins with areas greater than about 258 km 2 (100 sq miles), and we extended it to the smaller drainage basins for the purpose of this study (Fig. 2). This extension is somewhat problematic, because it leaves out three estimations of flood peaks above the curve; Bronco Creek near Wikieup, Arizona; Eldorado Canyon at Nelson Landing, Nevada; and Dragoon Wash at St. David, Arizona (Fig. 2 numbers 4 5, and 6 respectively). To include these three floods at or below the envelope curve will demand a major discontinuity in the shape of the curve. In 1987, Costa suggested that any flood estimate that falls above the envelope curve should be carefully reexamined. These three specific estimations were challenged earlier by several authors (see Enzel et al. 1993, who also summarized critiques by Carmody, 1980 and Malvick, 1980). More recently, House & Pearthree (1995) demonstrated that the published Bronco Creek flood magnitude is a large overestimation. Therefore, we accepted the extension of the US Bureau of Reclamation's curve as is. Curve A in Fig. 2 encompasses all of the largest flood magnitudes in the United States as reported by Costa (1987). Curve B is defined by the palaeoflood data from the lower Colorado River basin (Table 1). A comparison between the United States curve and the gauged and historical data from the Colorado River basin indicates
128
Y. ENZEL E T A L . 105 t|oodS in the U S A.maximum
104
........................................
i r
o
i
,o3
.......... .....
E
-
3C~
11
i i
liEs
Es
, Vi
................ "
O _o~\e
=n_E=
i
...........
.... i ....................................
~' :~
::
i
O -
102 ---....................~ -
i 9
7
.....................................
Envelopecurvecontrolling wr9 Paleoflood(see Fig.1 for location)
9 Pr
O
UsE Ca '
floods
l0 l ................................................................................................................................... B-forEnVelpal~e176 ofloentOdSgagedt ireCUrveS: he~andU.S.historic(COst 1987) a,data I 1
t
1
i I [Ill[
1
10
i I Ill
I
102
I Illll[
i
[ I ill!l
103
I
104
i I Illll
t
105
i
I IIII
106
Drainage Area (km2) Fig. 2. Envelope curves of the largest floods in the United States (curve A; from Costa 1987), the documented palaeofloods from Table 1 (curve B), and the largest gauged and historical floods (curve C; solid part from US Bureau of Reclamation (1990) and dashed part is our extension of the curve to drainage basins with smaller areas). The numbers 1 2, and 3 next to the three envelope-curve-controlling floods indicate the three floods discussed in text. that drainage basins in the Colorado River basin produce systematically smaller floods than drainage basins in other regions of the United States. Although we are concerned about the accuracy of the modern discharge estimates, we directed our efforts at identifying the general position and trend of the upper bound and its relation to the palaeoflood data. Therefore, curve C (Fig. 2) encompasses all of the floods except those which are obviously controversial discharge estimates. This curve can be used as a tool to identify those floods which warrant a reexamination, similar to the suggestion by Costa (1987), and the practice by Carmody (1980) and Malvick (1980). It is demonstrated in Fig. 2 that in the Colorado River basin, a substantial increase in the temporal scale of flood records does not change the position of the envelope curve based on the gauged and historical data. The palaeoflood discharges fall on or below the curve which envelopes the largest gauged and historical floods (Fig. 2). Although these palaeoflood discharges represent much longer periods of record and are usually larger than modern floods in the individual rivers where they were studied, they nevertheless are remarkably similar to the magnitudes of the largest modern or historical floods in the region. The relation between palaeoflood discharges and the regional envelope curve for the Colorado River basin is consistent with the concept that there is a physical or hydrometeorological limit on the magnitude of the maximum flood that can be expected in a given drainage (Costa 1987).
129
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
Because no other palaeoflood data base similar to the one for the lower Colorado River basin exists, the only way to further test the emerging pattern is on a subset of the data on a somewhat smaller and more hydroclimatically homogenous area. We chose the southern Arizona subregion because of data availability. The maximum floods that occurred in drainage basins within arid and semiarid southern Arizona are shown in Fig. 3. To produce a hydroclimatologically homogeneous data set we excluded from this figure gauging stations on the main stem of the Gila River, basins that drain to the Gila River from the north and have headwaters in high elevations, and one station that is clearly affected by urbanization in Tucson, Arizona. The resulting curve for the southern Arizona subregion is slightly different from curve C for the entire Colorado River basin. Smaller basins (0.2-1 km 2) were included in the curve for southern Arizona because the palaeoflood data for the smaller sized basins are exclusively from that region. Palaeoflood magnitudes were estimated for five small drainage basins in the Tortolita Mountains north of Tucson, Arizona, for one basin the White Tank Mountains, for Tiger Wash near the Harquahala Mountains, and for one basin in Sierra Estrella west of Phoenix, Arizona (Fig. 1, Table 1; House 1991; House et al. 1991; P. K. House, unpublished data, CH2MHill & French 1992). Although only one radiocarbon date is available for these palaeofloods, field evidence and relative age dating indicate that they are the largest floods to have occurred in these ungauged basins during at least the last several hundred years (Baker et al. 1990; House 1991). All of these palaeoflood discharges plot on or below the envelope curve constructed from the gauged data from southern Arizona (Fig. 3). MartinezGoytre et al. (1994) recently provided an additional eight maximum palaeoflood discharge estimations from southern Arizona. None is higher than the regional envelope curve shown in Fig. 3. In addition, none of the large floods which occurred in Arizona in January and February 1993 (several were the largest on record), 104
AA
,'-', 103
b
rw~U, R
,-- 102
'01
AA
SEi, t0
A Paleoflood magnitudes /~ From Martinez-Goytre et al (1994)
1
10 -l
I
I
I I
IIII
I
1
I
I IIIIII
I
10
I
I IIIIII
I
I
I IIIIII
102
I
103
I
I IIIIII
I
104
I
105
Drainage Area (km2) Fig. 3. Envelope curve for southern Arizona floods and palaeofloods from that region (letters are in reference to Table 1). Open triangles are the largest paleofloods in basins draining the Santa Catalina mountains (SC in Table 1) in Arizona estimated by Martinez-Goytre et al. (1994).
130
Y.
ETAL.
ENZEL
River basin
12
104 ~ ~
o
Lees ferry
-
~
A
-
~
B
103 ---
~
,.~ ~
_
~
-
.I--4
102
I
I
I
I IIIL
I
I
I
I
I IIIE
103
I
I
I
I
I
IIII
104 Drainage Area (km2)
Fig. 4. Envelope curves for: (A) the upper Colorado River drainage basin (all gauging stations in the basin from the headwaters to Green River, Utah and Cisco, Utah on the Green and Colorado Rivers, respectively; see inset in Fig. 1); (B) the upper and middle Colorado River basins (all gauging stations from the headwaters to Lees Ferry, Arizona; Fig. 1); (C) the entire Colorado River basin (similar to curve C in Fig. 2). exceeded the regional envelope curve (House 1993). The relationship between the palaeoflood discharges and the regional envelope curves is consistent with the assumption of an upper limit to flood magnitudes in both the entire Colorado River basin and the southern Arizona subregion. Analysis of other subregions within the Colorado River basin indicates that they have substantially different envelope curves for maximum instantaneous flood peaks (Fig. 4). Different types of storms produce the envelope-shaping peak floods in each subregion, indicating that hydroclimatology plays a major role in defining the curves. Although it is assumed by the authors that hydroclimatology is the cause for the upper limit of floods (Wolman & Costa 1984), additional inquiries into these phenomena are needed. The existence of a natural upper limit would raise questions about the basic assumption, intrinsic to frequently used models in the probabilistic approach of flood-frequency analyses, that the upper bound on flood magnitudes cannot be determined and that the largest floods are beyond the range of practical concerns. The patterns presented in Figs 2 and 3 indicate that the upper bound is not beyond that range.
Clustering of the largest floods To resolve the question of random versus clustering influences on the timing of maximum flood peaks we used evidence of 251 palaeofloods derived from palaeoflood chronologies on 19 rivers in the lower Colorado River basin (Fig. 1). The flood chronologies from these sites are listed, summarized, and discussed in detail by Ely (1992) and studies by Ely & Baker (1985), O'Connor et al. (1986a, b, 1994), Partridge & Baker (1987), Roberts (1987), Webb et al. (1988), Wohl (1989),
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
131
Melis (1990), Enzel et al. (1994) and Ely (in press). These chronologies are based on cultural artifacts and on more than 150 radiocarbon dates (listed in Ely 1992). The dated floods were placed into time intervals of 200 years each throughout the last 5600 radiocarbon years. These intervals are suitable to the resolution of the method and data. Unfortunately, there is no way to divide the record into smaller intervals. Combining the palaeoflood information into one data set allows a greater confidence in the emerging temporal patterns in the occurrence of large floods than would information based on single studies of individual rivers. The combined record (Fig. 5) of palaeofloods from the lower Colorado River area indicates a distinct pattern of large floods clustering in specific time episodes during the last 5000 years (Ely et al. 1993). Periods of numerous large floods span 5-3.6 ka, 2.2-0.8 ka and 0.6-0 ka. The episodes of few to no large floods are equally important from a climatic standpoint. The longest and most pronounced period of very few large floods is 3.6-2.2 ka. Another significant hiatus in large floods occurred from 0.8-0.6 ka, which shows a significant drop in the number of large floods immediately after an episode (1-0.8 ka) of particularly frequent high-magnitude floods. Periods of numerous large floods correspond to high lake levels (Mehringer & Warren 1976; Smith 1979; Enzel et al. 1989, 1992; Waters 1989; Stine 1990; Enzel & Wells in press) and groundwater discharge (Benson & Klieforth 1989) in the southwestern United States, global neoglacial advances (Rothlisberger 1986; Wigley 1988), and frequent strong E1 Nifio events (Anderson 1993) (Fig. 5). Enzel et al. (1989) and Enzel & Wells
Fig. 5. Composite chronology of the palaeofloods in the lower Colorado River basin in Arizona and southern Utah. The numbers of palaeofloods are arranged according to 200-year intervals. All ages are uncalibrated 14C dates. See text for references of the other palaeohydrological and palaeoclimatological records. Modified from Ely et al. (1993).
132
Y. E N Z E L E T AL.
(in press) have demonstrated that the formation and maintaining of short duration Holocene lakes in the terminus of the Mojave River in southern California could have been caused only by a large increase in the frequency of floods with magnitudes similar to the largest observed in the modern record (Enzel 1992). The increase in flood frequency during the last 400-600 years is apparent from Fig. 5. This can be either due to a bias in the methodology of palaeoflood hydrology or a true phenomenon. To try to detect a preservation bias we omitted chronologies with a record shorter than 400 years and normalized the number of floods by the number of rivers which produced data for each interval. The results (Ely 1992) did not alter the basic observation of increased flooding in the last several centuries. Therefore, the resulting pattern is most likely not merely an artifact of differential preservation. Other palaeohydrological evidence also indicate an increase in annual runoff in the Southwest (D'Arrigo & Jacoby 1991) and channel incision (e.g., Cooke & Reeves 1976; Webb et al. 1991) in the late nineteenth-early twentieth centuries, which is consistent with increased flooding in the region during that time. We assume that the causes of the largest floods during the late Holocene are similar to those of the modern floods. If the modern, regional oceanic and atmospheric conditions necessary to produce the largest floods are anomalous and unique, then similar conditions must have occurred to produce the large flood peaks. Modern storms which produce the largest floods in the sites summarized here are mainly winter North Pacific frontal storms and late-summer and fall storms associated with Pacific tropical cyclones over northwestern Mexico in conjunction with mid-latitude low-pressure troughs (Ely et al. 1994). Local summer convective storms are significant flood-producing storms only in basins much smaller than those used to construct the palaeoflood chronology (Ely 1992). We have determined that during both winter and tropical storms associated with large floods in the study area, the atmospheric circulation shifts the storm track southward toward the southwestern United States. The winter storm track is shifted far to the south of its normal position (Enzel et al. 1989; Ely et al. 1994). An unusually low pressure anomaly off the coast of California, and a high pressure anomaly over the Aleutians or the Gulf of Alaska are characteristics of the winter floods (Ely et al. 1994). Similar atmospheric patterns caused large winter floods in southern California (Enzel et al. 1989; Enzel & Wells in press). Tropical-cyclone floods exhibit a similar low-pressure anomaly and an extended blocking highpressure anomaly in the central North Pacific in addition to the entrainment of an eastern Pacific tropical storm (Ely et al. 1993). On the longer, late Holocene time scale, periods of increased flooding are associated with distinct, persistent changes in regional climate, large-scale atmospheric circulation patterns (Enzel et al. 1992; Ely et al. 1993), and an increased frequency of strong E1 Nifio events as reflected both in the Nile record (Anderson 1993; Quinn 1993) (Fig. 5) and in the E1 Nifio-like warm sea surface temperatures off the coast of California (Pisias 1978). The variations in flood frequency over the last 1000 years agree exceptionally well with these records (Ely et al. 1994). Combining the upper boundary approach with the suggested climatic forcing on the occurrence of the largest floods reveals the potential effect of climatic variations on the magnitudes for the rarest flood events in the region. The influence of climatic variability on the occurrences of extreme floods has been recognized at several time scales, and mechanisms to explain this association have been suggested (Knox 1983;
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
133
Webb 1985; Baker 1987b; Enzel et al. 1989; Ely and Baker 1990; Ely 1992; Ely et al. 1994; Enzel 1992; Webb & Betancourt 1992). Climate has also varied over different temporal scales during the late Holocene (e.g., Bradley 1985), the period for which palaeoflood data are available. However, not one Colorado River tributary where a palaeoflood study has been performed has produced a flood with a magnitude greater than the flood expected from the envelope curve of the modern record. In addition, a hydrological model of the Holocene lakes in the Mojave River terminus (Enzel et al. 1989) showed that although the floods that produced these lakes were similar to the modern largest floods, they could not have been much larger than the largest observed in the modern and historical records (Enzel 1992). The increased frequencies of large floods during distinct time periods over the last several thousand years were not necessarily associated with greatly increased peak flood magnitudes.
Summary Combining palaeoflood data from as many rivers and streams as possible allows important patterns to emerge. These patterns are unseen from data sets limited in their spatial and temporal coverage. The palaeoflood data have provided evidence for the existence of an upper limit to peak flood magnitudes in a region. Our results suggest that this upper limit can be estimated from the combination of modern, historical, and palaeoflood data (Enzel et al. 1993). This concept has been discussed in previous studies, but lack of detailed, high-quality, long-term data prevented the discussions to extend beyond suggestions. We stress that not all the rivers in the region will experience floods equivalent to the maximum discharge magnitude delineated by the envelope curve. The largest flood within a specific basin in the region can be smaller than the curve, but will not exceed the curve. The palaeoflood data also provided a regional chronology of large floods which exhibited a distinct clustering of floods through time. Comparison with other palaeoclimatic and palaeohydrologic data indicates a strong climatic control on the temporal distribution of floods in the southwestern United States (Ely et al. 1993). These patterns hold major implications for the study of flood hydrology, in particular to the increased understanding of the relationship between climate variability and floods, and application of flood hydrology toward a better understanding of the maximum expected flood on a given river. This research demonstrates the applicability of palaeohydrological information parameters in testing concepts and assumptions that have long affected scientific thought in surface-water hydrology. Figures were modified and drawn by A. Altman, Geography Department, Hebrew University. Research was supported by NSF grant 8901430. This publication is contribution 35 of the Arizona Laboratory for Paleohydrological and Hydroclimatological Analysis (ALPHA), University of Arizona. Correspondence to Y. Enzel (e-mail:
[email protected]).
References ANDERSON,R. Y. 1993. Long-term changes in the frequency of occurrence of E1 Nifio events. In: DIAZ, H. F. & MARKGRAF,V. (eds) El Niffo--Historical and Paleoclimatic Aspects of the Southern Oscillation. Cambridge University Press, Cambridge, 193-200.
134
Y. ENZEL ETAL.
BAKER, V. R. 1987a. Paleoflood hydrology and extraordinary flood events. Journal of Hydrology, 96, 79-99. - - 1 9 8 7 b . Paleoflood hydrology and hydroclimatic change. In: The Influence of the Climate Change and Climatic Variability on the Hydrologic Regime and Water Resources. IAHS Publication, 168, 123-131. & KOCHEL, R. C. 1988. Flood sedimentation in bedrock fluvial systems. In: BAKER, V. R., KOCHEL, R. C. & PATTON, P. C. (eds) Flood Geomorphology. John Wiley, New York, 123-137. , DEMSEY, K. A., ELY, L. L., FULLER, J. E., HOUSE, P. K., O'CONNOR, J. E., ONKEN, J. A., PEARTHREE, P. A. & VINCENT, K. R. 1990. Application of geological information to Arizona flood hazard assessment. In: FRENCH, R. n. (ed.) Hydraulic~Hydrology of Arid Lands. Proceedings of the 1990 International Symposium, San Diego, California, American Society of Civil Engineers, New York, 621-626. BENSON, L. V. & KLIEFORTH, H. 1989. Stable isotopes in precipitation and ground water in Yucca Mountain region, southern Nevada: paleoclimatic implications. In: PETERSON, H. (ed.) Aspects of Climate Variability in the Pacific and the Western America. American Geophysical Union, Geophysical Monographs, 55, 41-59. BENSON, M. A. 1964. Factors affecting the occurrence of floods in the Southwest. US Geological Survey Water-Supply Papers, 1580-D, D1-D40. BRADLEY, R. S. 1985. Quaternary Paleoclimatology. Allen and Unwin, Boston. CARMODY,T. 1980. A critical examination of the 'largest'floods in Arizona: a study to advance the methodology of assessing the vulnerability of bridges to floods for the Arizona Department of Transportation. The Engineering Experiment Station, College of Engineering, University of Arizona, Tucson, General Reports, 1. CHZMHILL & FRENCH R. H. 1992. Alluvial fan data collection and monitoring study. Final report for the Flood Control District of Maricopa County, Phoenix, Arizona. COOKE, R. U. & REEVES, R. W. 1976. Arroyos and Environmental Change in the American Southwest. Clarendon Press, Oxford. COSTA, J. E. 1987. A comparison of the largest rainfall-runoff floods in the United States with the People's Republic of China and the world. Journal of Hydrology, 96, 101-115. CREAGER, W. P. 1939. Possible and probable future floods. Civil Engineering, 9, 668-670. CRIPPEN, J. R. 1982. Envelope curves for extreme flood events. Journal of Hydraulic Engineering, 108, 1208-1212. -& BUE, C. D. 1977. Maximum floodflows in the conterminous United States. US Geological Survey Water-Supply Papers, 1887. D'ARRIGO, R. D. & JACOBY, G. C. 1991. A 1000-year record of winter precipitation from northwestern New Mexico, USA: a reconstruction from tree-ring and its relation to E1 Nifio and the South Oscillation. The Holocene, 1, 95-101. DOOGE, J. C. I. 1986. Looking for hydrological laws. Water Resources Research, 22, 46S-58S. ELY, L. L. 1992. Large floods in the southwestern United States in relation to late-Holocene climatic variations. PhD Dissertation, Department of Geosciences, University of Arizona, Tucson. in press. Response of large floods in the southwestern United States to late-Holocene climatic variations. Geomorphology. & BAKER, V. R. 1985. Reconstructing paleoflood hydrology with slackwater deposits, Verde River, Arizona. Physical Geography, 6, 103-126. & - - 1 9 9 0 . Large floods and climate change in the southwestern Untied States. In: FRENCH, R. H. (ed.) Hydraulics~Hydrology of Arid Lands. Proceedings of the 1990 International Symposium, San Diego, California, American Society of Civil Engineers, New York, 361-366. - - , O'CONNOR, J. E. & BAKER, V. R. 1988. Paleoflood hydrology of the Salt and Verde Rivers, central Arizona. Proceedings of the 8th Annual US Committee on Large Dams Lecture Series, Phoenix, Arizona, 000-000. --, ENZEL, Y., BAKER, V. R. & CAYAN, D. R. 1993. 5000-year record of extreme paleofloods and associated atmospheric circulation patterns, southwestern United States. Science, 262, 410-412. -
-
-
-
-
-
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
135
, & 1994. Anomalous North Pacific atmospheric circulation and large winter floods in the southwestern Untied States. Journal of Climate, 7, 977-987. ENZEL, Y. 1992. Flood frequency of the Mojave River and the formation of late Holocene lakes, southern California, USA. The Holocene, 2, 11-18. & WELLS, S. G. in press. Holocene paleohydrology and paleoclimatology from modern records of extreme events: example from southern California. Geomorphology. , CAYAN, D. R., ANDERSON, R. Y. & WELLS, S. G. 1989. Atmospheric circulation during Holocene lake stands in the Mojave Desert: evidence of regional climate change. Nature, 341, 44-47. , BROWN, W. J., ANDERSON, R. Y., MCFADDEN, L. D. & WELLS, S. G. 1992. Shortduration Holocene lakes in the Mojave River drainage basin, southern California. Quaternary Research, 38, 60-73. , ELY, L. L., HOUSE, P. K., BAKER, V. R. & WEBB, R. H. 1993. Paleoflood evidence for a natural upper bound to flood magnitudes in the Colorado River basin. Water Resources Research, 29, 2287-2297. - - , - - , MARTINEZ-GOYTRE, J. & VIVIAN, R. G. 1994. Paleofloods and a dam failure flood on the Virgin River, Utah and Arizona. Journal of Hydrology, 153, 291-315. FULLER, J. E. 1987. Paleoflood hydrology of the alluvial Salt River, Tempe, Arizona. MSc Thesis, Department of Geosciences, University of Arizona, Tucson. GEORGIADI, A. G. 1979. Upper limits of the elements of the hydrologic regime. Soviet Hydrology, 18, 225-230. HORTON, R. E. 1936. Hydrologic conditions as affecting the results of the application of methods of frequency analysis to flood records. US Geological Survey Water-Supply Papers, 771,433-449. HOUSE, P. K. 1991. Paleoflood Hydrology of the Principal Canyons of the Southern Tortolita Mountains, Southeastern Arizona. Arizona Geological Survey Open-File Reports, 91-6. - - 1 9 9 3 . The Arizona floods of January and February 1993. Arizona Geology, 23, 1-9. & PEARTHREE, P. A. 1995. A geomorphologic and hydrologic evaluation of an extraordinary flood discharge estimate: Bronco Creek, Arizona. Water Resources Research, 31, 3059-3073. , & VINCENT, K. R. 1991. Flow patterns, flow hydraulics, and flood-hazard implications of a recent extreme alluvial-fan flood in southern Arizona. Geological Society of America Abstracts with Programs, 23, A-121. KNOX, J. C. 1983. Responses of river systems to Holocene climates. In: WRIGHT,H. E. (ed.) Late Quaternary Environments of the United States--The Holocene. University of Minnesota Press, Minneapolis, 26-41. - - 1 9 9 3 . Large increase in flood magnitude in response to modest changes in climate. Nature, 361, 430-432. KOCHEL, R. C. 1980. Interpretation of flood paleohydrology using slackwater deposits, lower Pecos and Devils Rivers, southwestern Texas. PhD Dissertation, University of Texas, Austin. -& BAKER, V. R. 1982. Paleoflood hydrology. Science, 215, 353-361. & RITTER, D. F. 1987. Implications of flume experiments for the interpretation of slackwater paleoflood sediments. In: SINGH, V. P. (ed.) Regional Flood Frequency Analysis. D. Reidel, Boston, 371-390. , BAKER, V. R. & PATTON, P. C. 1982. Paleohydrology of southwestern Texas. Water Resources Research, 18, 1165-1183. MALVICK, m. J. 1980. A magnitude-frequency-area relation forfloods in Arizona. A study to
-
-
-
-
-
-
advance the methodology of assessing the vulnerability of bridges to floods for the Arizona Department of Transportation. The Engineering Experiment Station, College of Engineering, University of Arizona, Tucson, General Reports, 2. MARTINEZ-GOYTRE, J., HOUSE, P. K. & BAKER, V. R. 1994. Spatial variabilities of smallbasin paleoflood magnitude for southwestern Arizona mountain range. Water Resources Research, 30, 1491-1501. MATTHAI, H. F. 1969. Floods of June 1965 in South Platte River basin, Colorado. US Geological Survey Water-Supply Papers, 1850-B, B1-B64.
136
Y. ENZEL ETAL.
MELIS, T. S. 1990. Evaluation of flood hydrology on twelve drainages in the central highlands of Arizona: an integrated approach. MSc Thesis, Department of Geology, Northern Arizona University, Flagstaff. MEHRINGER, P. J. JR. & WARREN, C. N. 1976. Marsh, dune and archeological chronology, Ash Meadows, Amargosa Desert, Nevada. In: ELSTON, R. & HEADRICK, P. (eds) Holocene Environmental Change in the Great Basin. Nevada Archeological Survey Research Papers, 6, 120-150. O'CONNOR, J. E. & WEBB, R. H. 1988. Hydraulic modeling for paleoflood analysis. In: BAKER, V. R., KOCHEL, R. C. & PATRON, P. C. (eds) Flood Geomorphology. John Wiley, New York, 383-402. ---, -& BAKER, V. R. 1986a. Paleohydrology of pool and rime pattern development, Boulder Creek, Utah. Geological Society of America Bulletin, 97, 410-420. - - , FULLER, J. E. & BAKER, V. R. 1986b. Late Holocene flooding within the Salt River basin, central Arizona. (ALPHA) Dept. Geosciences, Univ. Arizona, Tucson, Arizona. , ELY, L. L., WOHL, E. E., STEVENS, L. E., MELIS, T. S., KALE, V. S. & BAKER, V. R. 1994. A 4500-year record of large floods on the Colorado River in the Grand Canyon, Arizona. Journal of Geology, 102, 1-9. PARTRIDGE, J. B. & BAKER, V. R. 1987. Paleoflood hydrology of the Salt River, Arizona. Earth Surface Processes and Landforms, 12, 109-125. PISIAS, N. G. 1978. Paleoceanography of the Santa Barbara basin during the last 8000 years. Quaternary Research, 10, 366-384. QUINN, W. H. 1993. A study of Southern Oscillation-related climatic activity for, A.D. 622-1990 incorporating Nile River flood data. In: DIAZ, H. F. & MARKGRAF, V. (eds) El NiJqo--Historical and Paleoclimatic Aspects of the Southern Oscillation. Cambridge University Press, Cambridge, 119-150. ROBERTS, L. K. 1987. Paleohydrologic reconstruction, hydraulics, and frequency-magnitude relationships of large flood events along Aravaipa Creek, Arizona. MSc Thesis, Department of Geosciences, University of Arizona, Tucson. ROESKE, R. K. 1978. Methods for estimating the magnitude and frequency of floods in Arizona. Report ADOT-RS-15(121): Prepared for the Arizona Department of Transportation by US Geological Survey, Water Resources Division, Tucson, Arizona. ROTHLISBERGER, F. 1986. 10,000 Jahre Gletschergeschichte Der Erde. Verlag Sauerlander, Aarau, Switzerland. SMITH, G. I. 1979. Subsurface stratigraphy and geochemistry of late Quaternary evaporites, Searles Lake, California. US Geological Survey Professional Papers, 1043. SMITH, S. S. 1990. Relationship of large floods and rapid entrenchment, Kanab Creek, southern Utah. MSc Thesis, Department of Geosciences, University of Arizona, Tucson. STINE, S. 1990. Late Holocene fluctuations of Mono Lake, eastern California. Palaeogeography, Palaeoclimatology, Palaeoecology, 78, 333-381. US BUREAU OF RECLAMATION 1990. Colorado River Basin--Probable Maximum Floods, Hoover and Glen Canyon Dams. US Department of Interior, Washington DC. WATERS, M. R. 1989. Late Quaternary lacustrine history and paleoclimatic significance of pluvial Lake Cochise, southeastern Arizona. Quaternary Research, 32, 1-12. WEBB, R. H. 1985. Late Holocene floods on the Escalante River, south-central Utah. PhD Dissertation, Department of Geosciences, University of Arizona, Tucson. & BETANCOURT, J. L. 1992. Climatic variability andfloodfrequency of the Santa Cruz River, Pima County, Arizona. US Geological Survey Water-Supply Papers, 2379. --, O'CONNOR, J. E. & BAKER, V. R. 1988. Paleohydrologic reconstruction of flood frequency on the Escalante River, south-central Utah. In: BAKER, V. R., KOCHEL, R. C., & PATTON, P. C. (eds) Flood Geomorphology. John Wiley, New York, 402-418. , SMITH, S. S. & MCCORD, V. A. S. 1991. Historic Channel Change of Kanab Creek, Southern Utah and Northern Arizona. Grand Canyon Natural History Association Monographs, 9. WIGLEY, T. 1988. The climate of the past 10,000 years and the role of the Sun. In: STEPHENSON, F. & WOLFENDALE,A. (eds) Secular Solar and Geomagnetic Variations in the Last 10,000 Years. Kluwer Academic Publishers, Dordrecht, 209-224.
MAGNITUDE & FREQUENCY OF HOLOCENE PALAEOFLOODS
137
WOHL, E. E. 1989. Large floods in Redfield Canyon. Final Report for contract no. IGA-896125-210-0111, Arizona Department of Water Resources.
WOLMAN, M. G. t~ COSTA, J. E. 1984. Envelope curves for extreme flood events, Discussion. Journal of Hydraulic Engineering, 110, 77-78. YEVJEVlCH, V. M. & HARMANCIOGLU, N. B. 1987. Research needs in flood characteristics. In: SINGH, V. P. (ed.) Application of Frequency and Risk in Water Resources. D. Reidel, Boston, 1-21.
From Branson, J., Brown, A. G. & Gregory, K. J. (eds), 1996, Global Continental Changes." the Context of Palaeohydrology, Geological Society Special Publication No. 115, pp. 139-153
The response of geomorphic systems to climatic and hydrological change during the Late Glacial and early Holocene in the humid and sub-humid tropics MICHAEL
F. T H O M A S
1 & MARTIN
B. T H O R P 2
1Department of Environmental Science, University of Stirling, Stirling FK9 4LA, UK. 2 Martin B. Thorp, Department of Geography, University College Dublin, Belfield, Dublin 4 Republic of Ireland Abstract: Dated alluvial stratigraphies indicative of late Quaternary environmental change in the humid tropics have increased, but the database remains inadequate and the intensity and duration of wet-dry oscillations and responses of hillslopes and river systems remain poorly understood. Dry conditions at the Last Glacial Maximum were marked by semi-arid landforms with reduced stream activity. Large palaeofloods, valley-floor erosion, channel cutting and flood deposition occurred at the Pleistocene-Holocene transition after 13000BP. Distinctive floodplains and stratigraphies characterized by multiple shifts from lateral to vertical accretion were built initially over a period of nearly 2 ka after 9 500 BP during the early Holocene pluvial in Africa, Kalimantan and Amazonia during and after re-establishment of the lowland rainforests. Several wet-dry climatic oscillations followed in the mid-Holocene period and are marked by alluvial cut and fill sequences and by slightly thinner but coarse textured floodplain overbank sediments.
Late Pleistocene and Holocene signals of climate change are becoming more widely available for the tropics, mainly from studies of pollen and lake sediments. However, there is a gap in our knowledge between the inferences about vegetation and climate and our understanding of river behaviour and landscape dynamics as responses to the changes. That understanding needs to come, in part at least, from the analysis, dating and interpretation of alluvial deposits. Unfortunately, there is a dearth of published alluvial evidence from within the humid tropics (Thomas 1994). There are several problems associated with environmental and palaeohydrological interpretations of alluvial sediments. It is not always easy to distinguish the impacts of regional changes in controlling factors such as catchment relief, tectonics, and base level, from the effects of site sensitivity on the response to single high magnitude events or to secondary threshold effects (complex response) within the geomorphic system (Schumm 1977). Progressive stripping of relict late Tertiary weathering mantles during episodes of rapid landscape change must also have a cumulative impact and also limit the potential for repeated events at the same site. The impact of human occupation on both pollen spectra and on lake sedimentation (Flenley 1988) must also clearly lead to other uncertainties of interpretation, and research into the impacts of possible anthropogenic vegetation change, as early as 26 000 BP in Irian Jaya (Haberle et al. 1991) and occurring more widely by 7-5 ka BP, on landscape dynamics and river behaviour is at a primitive stage.
140
M. F. THOMAS & M. B. THORP
A second general problem facing palaeohydrological interpretation concerns the nature of past climate, particularly rainfall regimes, where close analogies with current climates may be inappropriate. The notion of geomorphologically effective rain events based on the concept of changing magnitude and frequency over time and space is arguably the most appropriate way forward in this field (Wolman & Miller 1960; Ahnert 1987; Ehbergen & Imeson 1989). Unfortunately, existing palaeoenvironmental data and models of palaeoclimate do not usually afford such a detailed understanding. It is now accepted that climate changes can be abrupt, occurring over 102 years rather than 103 years, and that particular climatic patterns may persist for a few hundred years only (Street-Perrott & Perrott 1990; Gasse & Van Campo 1994). Consequently low resolution studies of sediments and peats can be misleading. On the other hand, high-resolution studies will converge with concepts of flood frequency and inter-annual climate variability and event stratigraphy. Equally worrying are the doubts expressed concerning the supposed relationships between climate, vegetation and sediment yield on the one hand, and predictions about stream power, erosion and deposition on the other. Much of our thinking in this regard for regions outside the recently glaciated areas derives from work done in the temperate continental and the arid zone climates (Langbein & Schumm 1958; Schumm 1965, 1968; Knox 1972, 1984; Bull 1991). Thus it is common to indicate a peak of sediment production during periods of rapidly changing climate, as when rainfall increases ahead of vegetation recovery (Knox 1972, 1983; Thomas & Thorp 1980, 1992; Roberts & Baker 1993). In a comparable manner rivers are expected to aggrade their valleys when sediment yield is high but discharge is erratic, and to erode their beds when forested conditions promote broader flood peaks and lower sediment concentrations. However, not all of this reasoning is matched to field evidence and some simple models of these kinds may be very misleading. Several issues can be listed: (1) The model of long periods of 'stable' climate punctuated by intervals of rapid change is not adequate to account for landscape complexity, because it largely ignores the magnitude-frequency properties of climate; additionally, 'stable' climates may not have persisted throughout the periods indicated on most models. (2) The rainfall regime is critical to any arguments concerning stream activity. Runoff trends vary with storm size and antecedent moisture and can be almost continuous under equatorial conditions. In monsoonal climates, on the other hand, there is a period of major annual floods, creating a distinctive fluvial regime (Gupta & Dutt 1989), different from that of the drier seasonal savanna climates found in parts of Africa and S America. In his analysis of rainfall erosivity in tropical climates in West Africa Roose (1977, 1981) found immense variations across the region in the erosivity of rainfall such that slight spatial shifts in vegetation communities can induce major changes in sediment yield. His model is unlikely to apply in all tropical climates, but where the rainfall generating mechanism is similar throughout an area it may have great relevance to debates about landscape response to rainfall and vegetation changes. (3) The production of sediment from hillslopes may arise from mass movements favoured by saturated soil conditions, as well as by surface wash which is promoted by open vegetation. But in neither case are connections to stream channels always direct, and sediment stores in the landscape play a major role in controlling delivery into rivers, a point made forcibly by Church & Slaymaker (1989) with respect to glaciated terrain. Thus, while highland streams may respond to the magnitude and frequency of
HYDROLOGICAL CHANGE IN HUMID/SUB-HUMID TROPICS
141
landslides coming directly into the channel, plains rivers create floodplains over 104 years, and these form the major sources of sediment entering the channel. (4) Late Quaternary events were superimposed on landscapes produced during many previous cycles or oscillations of climate. Those landscapes contain glacis, fans, terraces and a multitude of hillslope forms, some parts of which act as barriers to the entry of sediment into present-day rivers and strongly influence morphostratigraphy of Holocene floodplains. It is clear, therefore, that appraisal of this group of problems requires a close scrutiny of the processes involved and of their periodicities and controlling factors (Eybergen & Imeson 1989). However, published knowledge of current catchment geomorphological behaviour in response to meteorological events is limited (see Thomas 1994 and Douglas & Spencer 1985 for collations of data). For these and other reasons, the interpretation of many alluvial forms and deposits must depend on site properties, their sensitivity to change and their regional settings. The analysis of material from vertical cores taken from enclosed depressions does not always satisfy this requirement, while the interpretation of lake levels is notoriously difficult, where basin morphology, catchment size, altitudinal variation and hydrogeology can all be extremely variable (Gasse & Van Campo 1994). However, the use of a large number of observations has allowed world-wide correlations to be offered for Late Quaternary lake levels (Street & Grove 1979; Street-Perrott et al. 1985), and this comparative method is important for other studies, including those of hillslope morphology and stream sedimentation. In this review the alluvial responses to environmental change in the humid tropics will be confined to the period spanning the recovery from the L G M through to
Table 1. Chronology of late Quaternary environmental change in the humid tropics Radiocarbon Years BP 3 100-2 400 3 400-3 100 4 200-3 400 5 500-4 200 7 000-5 500 7 800-7 000 10 500-8 000 11000-10 500 12 500-11000 13 000/12 000post-22 000
Probable environmental conditions (recorded examples in brackets) Possibly drier, accompanied by deforestation and human occupation, and continuing to the present day. West Africa (Ghana, post2400 years BP Brazil) Increased humidity in forested tropics with rising discharges, several lesser oscillations of humidity followed Mid Holocene dry phase, probably quite severe (Africa, Brazil) Declining humidity in some areas of humid tropics (dry excursions in Amazonia, 5500; 4800) Increased humidity and modest rise in lake levels Reduced lake levels and river discharges in W & E Africa, Brazil Second humid period with high lake levels and discharges; reestablishment of forest Dry, cool interval in many areas; low lake levels (Younger Dryas) Rapid warming with unstable climates and prolonged heavy rains in tropical Africa, very high lake levels (world-wide) Becoming cold in uplands and dry in most lowlands; by 18 000 treeline depressed 1000 m, rainfalls possibly reduced by 50% (most sites)
Main Sources: Absy et al. (1989, 1991); Adamson et al. (1980); Butzer 1980; Kershaw (1978, 1992); Schubert (1988); Street & Grove (1979); Talbot et al. (1984); Thomas & Thorp (1980, 1985); Thorp et al. (1990); Street-Perrott & Perrott (1990); Gasse & Van Campo (1994).
142
M. F. THOMAS & M. B. THORP
c. 3 ka BP, when the environmental signal in alluvial stratigraphies becomes strongly mixed with that from human activities. The chronological framework for this period is tentatively generalised and summarized in Table 1. Naturally, such a summary table hides many of the disagreements and differences between regions in the number, onset and duration of major climate changes. Distinguishing the effects of human environmental modification from those of climatic change is a further problem. Preliminary indications suggest that early deforestation in the lowland humid tropics c. 10-4 ka BP was probably small scale, short-term and spatially very patchy and its impacts may have been short-lived and localized, until widespread agriculture associated with larger populations spread across many tropical areas in a diachronous fashion, perhaps from about 5 ka BP onwards, but mainly during the last 2-3 ka (Flenley 1988; Hope & Tulip 1994; Jolley et al. 1994).
Conditions during the Late Glacial Maximum The duration of ice-age aridity varied from region to region appearing post 22 ka BP in most areas and, according to some writers, ameliorating by 16.5 ka BP in SE India, (Van Campo 1986), by 15-14 ka BP in the west Zaire basin (Kadomura 1995) whilst elsewhere, dry climates seem to have persisted until well after 14 ka BP. Evidence of geomorphological conditions during the Late Glacial Maximum throughout the present humid and seasonally humid tropics includes: 9
9
9 9 9 9 9 9
deep sand-filled desiccation cracks in the weathered phyllite beneath shallow tributary valleys floors now buried beneath early Holocene colluvial and alluvial gravels in southern Ghana (Junner 1943; Hall et al. 1985); an apparent absence of alluvial sedimentary units in both headwaters and trunk streams between 25 ka BP and 13 ka BP in Amazonia (Van der Hammen et al. 1992a & b), West Africa (Thomas & Thorp 1980, 1985; Hall et al. 1985) and Kalimantan (Thorp et al. 1990, Thorp & Thomas, 1992); dry lake beds and swamps; regolith-stripped slopes found widely under present day forest and other woodland areas; stonelines within the forest and savanna zones; palaeopans, lunettes and other dune sands in Venezuela (Tricart 1982, 1985); large fans of course material as in the Pantanal (Mato Grosso, Brazil) (Klammer 1982), the Orinoco Llanos (Clapperton 1993a); widespread colluvium (Thomas 1994).
Collectively these indicate relative aridity and a suppression of the characteristic fluvial processes which justify suggestions of rainfall reductions of 30-66% compared to present values for large areas of the lowland tropics (RossignolStrick & Duzer 1979; Verstappen 1980; Peters & Tetzlaff 1990; Crowley & North 1991; Heaney 1991: Thomas 1994; Van der Hammen & Asby 1994). These conclusions pose questions regarding the vegetation cover at the LGM which are impossible to answer on the scale of the global tropics. In mountain areas, reduced temperatures depressed treelines by c. 1 km, but the impact of these changes on the
HYDROLOGICAL CHANGE IN HUMID/SUB-HUMID TROPICS
143
lowlands remains uncertain. The appearance of montaine flora in lowland pollen spectra from west Africa (Maley 1991; Giresse et al. 1994) and Panama (Bush et al. 1992) has been used as an argument against the refugia theory of Haffer (1969, 1987), but in terms of landscape ecology and dynamics this may be misleading. Forested conditions clearly persisted throughout the last glacial cycle in certain favoured areas that may have included western equatorial Africa (Preuss 1990; Maley 1991) and western Amazonia (Van der Hammen & Absy, 1994). But conditions were probably dry enough after c. 20 ka BP for 5-7 Ma for many areas of former forest to become mosaics of open woodland and grassland (Giresse et al. 1994; Van der Hammen & Absy 1994). Arguments about biodiversity in relation to so-called refugia are clearly contentious, but the persistence of lowland TRF during the last glacial cooling ( - 4 ~ to 6~ must have occurred in climatically and edaphicaly favoured (refuge or heartland) areas, albeit in modified form. In any case major reductions in rainfalls (30 to 60%) led to the reduction of water surpluses and dwindling of runoff and streamflow (Thomas & Thorp 1980; Van der Hammen & Absy 1994) beneath the prevailing vegetation cover. Towards the margins of the humid tropics, it seems likely that strengthened Trade winds, less frequent storm activity and reduced rainfall could have combined to produce landscapes of very open woodland.
The late Pleistocene-Holocene transition The onset of the late Pleistocene-Holocene transitional period is of great interest geomorphologically in the humid tropics and there is evidence for a period of unstable and often very wet conditions beginning around 12.5kasP and lasting e. 1.5 ka. It is remarkable how frequently this date is referenced around the globe. In Africa, wetter conditions appear to have become established near the Equator by 13 ka SP but did not become effective in the southern margins of the Sahara until 10 or even 9 ka BP (Fabre & Petit-Maire 1988). This illustrates the time-transgressive nature of climate changes when viewed across regions such as west Africa (Fairbridge 1976). It may also mean that the number of 'effective' fluctuations in rainfall will have varied significantly between areas, being fewer in the more humid tropics, where the forests probably survived drier episodes except in unfavourable edaphic sites. Large palaeofloods on the Nile and the Niger and also in smaller catchments in the seasonally humid tropics of west Africa commenced after 12.7 ka BP. The overflow of Lake Victoria and the deposition of the Sheik Hassan silts 30 m above the floodplain of the Nile at Wadi Haifa (the Wild Nile) have been given dates of 12.5-11.5 ka Be (Paulissen 1989). The first major depositional events of this period in the middle Birim River of Ghana and in the Bafi-Sewa headwaters in Sierra Leone date to the same period (12.7-12.4kasp) and are associated with extensive erosion of valley floors (Thomas & Thorp 1980; Hall et al. 1985; Thorp & Thomas 1992). On the other hand, the major pulse of off-shore sedimentation, at least from major African rivers, appears to have taken place after 12 ka BP, with the sedimentation peaks in the Niger delta dated to c. l l . 5 k a s p (Pastouret et al. 1978) and in the Congo estuary to 11.23 ka BP (Giresse & Lanfranchi 1984), whilst the lower sapropel muds of the eastern Mediterranean are bracketed between l l.76kaBP and 10.44kaBP (Rossignol-Strick et al. 1982). The sedimentation rates increased by four times in the Congo estuary and by 18 times in the Niger delta at these times. It is
144
M . F . THOMAS & M. B. THORP
tempting to suggest that the lag between fluvial erosion in the interior and the peak in offshore delta sedimentation reflects the sediment transfer time between the two, involving sediment storage on lower slopes and in floodplains. After 13 ka BP, hillslopes were probably subjected to more frequent and powerful sheetflood events and tributary channel extension led to the evacuation of colluvium from hillslope hollows. A fluvial landscape reasserted itself, by reactivating drainage lines. Prolonged rain, falling on regolith with high-rainfall acceptance, could also have led to landsliding, and to continued widespread fan formation especially in modern savanna areas. These fans were later abandoned or trenched as the sediment supply from upstream sources diminished and regular stream flows, albeit seasonal, became established. During the very large flood events which appear to have occurred during this phase, the critical power of streams appears to have been enough to remove coarse, often boulder sized, channel sediments, leading to erosion of 'bedrock' channels in Sierra Leone and Ghana. In eastern Zambia modern 'dambos' draining hillfoot zones have indistinct or no stream channels today, but exhibit gravel terraces 2-3 m above the valley floors, and also contain 3 m of fine gravel and sandy sediment infilling buried channels. Towards the end of this transition period, and following the Younger Dryas, with increasing vegetation cover and soil development, it may be surmised that flows became more seasonally regular and that flood peaks on many lowland tropical rivers probably became broader and less high. This would have allowed rivers to build alluvial plains and in due course to become single thread meandering streams. There are, however, few dated records of floodplain overbank alluvia from this period.
The early Holocene Pluvial Following the dry-cool Younger Dryas episode, the subsequent development of pluvial forested conditions, marked by high lake levels and by fluvial sedimentation, is clearly attested in the environmental records for which there are abundant early Holocene (mostly post-10 ka BP) dates. What is not clear, however, is how long these conditions persisted, neither is it agreed what rainfall mechanisms were predominant and therefore what the magnitude and frequency of rainfall events may have been. However, L6zine & Vergnard-Grazzini (1993) have suggested a rainfall 300mm wetter than today for the west African tropics at this time, a figure also used by Kershaw & Nix (1989) for northeast Queensland (although they recognized that the figure could be as high as 800 mm). This interglacial pluvial is sometimes represented as persisting well into the middle Holocene, but after c. 8 ka BP there is evidence of drier conditions lasting perhaps 500 years in Ethiopia, western Sahara and Ghana (Gasse & Van Campo 1994; Talbot & Johannessen 1992), with a return to greater humidity of climate after c. 7 ka BP. A severe dry phase followed in Africa after 5 kaBP and Van der Hammen et al. (1992a, b) indicates three dry phases between 5 and 3 ka BP in Colombian Amazonia. Thus whilst there is considerable spatial synchroneity around the 11-10 ka BP and the 4-3 kaBP cooler and drier periods there is considerable spatial variation in their onset and termination and as more detail becomes available, the division of the Holocene climates into longer periods of climatic stability separated by shorter periods of rapid change may become less tenable.
HYDROLOGICAL CHANGE IN HUMID/SUB-HUMID TROPICS
145
Records from W Africa, Australia and Brazil suggest the re-establishment of lowland forest between 9 and 8kaBP (Talbot & Delibrias 1980; Markgraf 1989; Servant et al. 1989; Kadomura & Maley 1994), in contrast to the absence of records for forest cover during the Pleistocene-Holocene transition. It is tempting, therefore, to suggest that in many areas the forest taxa were no longer present locally, and that migration of the rainforest from refuge areas such as western Amazonia and western equatorial Africa took up much of the period and was arrested for 500-1000 years during the Younger Dryas. The environmental controls over this recovery may have included the unstable climate of the late Pleistocene and earliest Holocene, and also the absence or continued loss of soil cover in many areas. It has been hypothesised that the period in itself was one of rapid hillslope erosion and landscape change (Thomas & Thorp 1980, 1992; Hall et al. 1985; Roberts & Baker 1993), and this must have opposed the re-establishment of equilibria (biostasie) in landscapes that had experienced 6-7 ka of dry conditions. The development of maximum pluvial conditions after the Younger Dryas event is recognised not only in the second Mediterranean sapropel, which accumulated in the period from c. 9-8 kaBP, but is clearly recognized in the several published dated alluvial stratigraphies. In West Africa, in the Birim River of Ghana, early Holocene floodplain sedimentation occurred after 9 ka BP and continued for more than 1000 years during which time several 2-3 m gravel units were formed together with a final 5-7 m thick fine member overbank unit. This records a major aggradation of the valley floor, burying 5 m thick coarse gravels which had already accumulated during the Late Pleistocene transition period between 13 and 12 ka BP within a deep bedrock channel incised into the pre LGM, late Pleistocene bedrock surface (Hall et al. 1985). The overbank fine members of the early Holocene floodplain are dominated by clays, silts and fine sands in contrast to more sandy textured later Holocene floodplains. In Sierra Leone, the Bafi-Sewa headwaters exhibit a similar time concentration of channel sediments between 9.5 and 7.8 ka BP, and their overbank fine members also contain organic rich clay facies within generally finer textured sediments than those of the later Holocene inset floodplains. In equatorial western Kalimantan, floodplain sedimentation also commenced around this time with dates from basal gravels beneath the floodplains of the Mandor and Raya rivers returning dates of 9.97 kaBP and 10.25 ka BP respectively (Thorp et al. 1990; Thorp & Thomas 1992). In their recent study of Holocene palaeochannels of the Yom river in the central plain of Thailand, Bishop & Godley (1994) identify several Holocene channel and floodplain formations. The early Holocene floodplain appears to have been constructed by higher discharges than the later units. Their calculations indicate a reduction factor of x3-4 for bankfull discharge (Qbf) between the early and the middle Holocene discharges. The greater humidity of the early Holocene is attested also by Lrffler et al. (1984) and by Nutalaya et al. (1989) as quoted by Bishop & Godley (1994). The Holocene sands and clays of the Blue (B) and White (W) Niles also show comparable flood peaks in the early/mid-Holocene, according to Williams (1980); Adamson et al. (1980): 8.4-8.1 kaBP (W); 7.SkaBP (B); cTkaBP (B & W). But the picture is less clear in the late Holocene. High flows on the Congo/Zaire River dating to the period 10-8 ka BP are referred to by Preuss (1990). In the Caquetfi valley (Colombia), late Pleistocene to early Holocene floodplain formation commenced after c. 12.6 ka BP and persisted until around 11 ka BP, after
146
M. F. THOMAS & M. B. THORP
which the sequence was partly eroded during the Younger Dryas/E1 Abra stadial (Van der Hammen et al. 1992a, b). After 10 ka BP they record renewed sedimentation and 'possible extensive inundation' which has again been evident in the last 3000 years due, they think, to Andean deforestation. Results from some of these rivers are summarized in Table 2. During the period from c. 10.25 to 8 ka BP forested condition became widespread (Lrzine &VergnaudGrazzini, 1993). Sediment supply was initially high with a wide range of textural sizes but may have declined to fine textured suspension loads with forest closure. Although deep-seated landslides are common in the humid tropics today, there is also abundant evidence of palaeolandslides in hilly terrains within savanna areas. Discharges appear to have been greater than today and single thread rivers built thick overbank depositional sequences. Stabilization of stream banks by fine silts and by vegetation would have inhibited lateral migration and irregularly sinuous, box-shaped channels, as in the Birim, became common on many rivers. The compromise between vertical accretion and overbank flooding meant that the latter became rarer, and occasional (102 year) high floods probably led to avulsion and splay formation on floodplains and to the evacuation of sediments from the small valleys, including the dambos and bolis. During and after the mid-Holocene, possibly from as early as 7.8kaBP and certainly after 5 ka BP drier conditions periodically recurred possibly lasting c. 500 years, but the forest remained largely intact although after 5kaBP it became increasingly subject to agricultural burning and clearance. It is not clear what the impact of these drier periods was on the rivers of the time, except that they must have experienced diminished flows within channels designed for much larger discharges; possibly nothing much happened. However, bank instability and cavitation may have occurred, and during subsequent high floods, some degree of lateral stream migration, together with the generation of large amounts of fine sediment from bank erosion would have characterised the system. After c. 5 ka BP alluvial stratigraphies record several periods of enhanced floodplain erosion and rebuilding, supra-basal gravel reactivation and minor cut and fill assemblages. Of interest are the erosional changes which terminated the major sedimentation periods. Those between the Transition and the early Holocene pluvial are perhaps Table 2. Dates of late Quaternary sedimentation in Blue (B) and White (W) Nile, Caquet6 (Colombia), Koidu Basin (Sierra Leone) and Birim (Ghana) in
Nile*
Caquetfit
C 14
Birim~
years Be
Koiduw
>2090- < 1780 c. 3000-c. 2300 >4300-c. 3200 7 000 (B & W) 7 500 (B) 8 400-8100 (W) 12 500-11 000 (B & W)
10 000-8800-7500 > 9500-7800 12 600-11 000 > 12 700- < 12 400-12 500- < 10 500
* After Williams (1980); Adamson et al. (1980). t After Van de Hammen et al (1992a, b). Ater Hall et al. (1985). wAfter Thomas & Thorp (1980).
HYDROLOGICAL CHANGE IN HUMID/SUB-HUMID TROPICS
147
more easy to explain than the erosion between the early Holocene floodplains and those built during the last 5000 years. Ad hoe hypotheses might be advanced invoking complex response or externally driven changes in water and sediment discharges after c. 7 ka BP, or the effects of high magnitude flow events during the several periods of reduced overall precipitation. However, as Bull (1991) has commented, rivers experiencing long term degradation due to tectonic factors, erode their beds whenever conditions for rapid sedimentation are absent. This wider view probably affects many of the river reaches quoted in this study.
Concluding discussion At the Pleistocene-Holocene transition climates are thought to have become unstable, but the changes to rainfall amount and distribution are little understood. Widespread evidence for very high floods on rivers both large and small suggests a major increase in storm size and frequency in low latitudes impacting on landsurfaces partially adjusted during the preceding 5000-9000 dry years to more arid conditions, while the pan-tropical rise in lake levels indicates a major change in water balance, reflecting enhanced rainfall totals rather than reduced evaporation. However, while this transition can be explained largely by insolation forcing, subsequent Holocene fluctations in tropical rainfall cannot, and variations in sea surface temperatures, or ocean currents, also fail to offer an explanation. Gasse & Van Campo (1994) have recently proposed that feedback mechanisms associated with the land surface conditions themselves may have been largely responsible. With initial post glacial warming and recovery of the monsoon rainfall mechanisms, came the expansion of forest vegetation, wetlands and lakes, decreasing albedo and increasing methane production into the atmosphere, reinforcing the insolation mechanism. The evaporation of the moisture in soils and lakes would subsequently require large amounts of solar energy and lead to a cooling of the surface, reversing the rainfall trends, and according to these authors, leading to a drying out of the land and a repetition of the cycle of rainfall fluctuation. They also point out that the declining magnitude of the fluctations follows the curve of N hemisphere insolation during the Holocene. High runoff and sediment yields would have characterized many hillslopes at the time of the first major increase in rainfall and frequent debris flows combined with fan building and trenching may have taken place. Many 'head' deposits and former valley fills in hilly terrain may have been excavated, but there is evidence from the dambo valleys of the Nyika Plateau in Malawi (Meadows 1985) and from the Inyanga Highlands of Zimbabwe (Tomlinson 1974) to suggest a period of rapid sedimentation post 12 ka BP. It also seems likely that deeper seated landslides would have been generated as climates became increasingly humid. Rivers were subject to very high flood peaks which combined to create major fluxes of water and sediment in large catchments. There may be an important association here between rapid hillslope erosion, especially from earlier sediment stores, deposition of coarse sediments, the flushing of clays through the catchment system and the marked accumulation of kaolinitic clay in deltas and estuaries. The Younger Dryas was probably important in the tropics as a period of drier climates lasting more than 500 years and this must have halted the spread of the
148
M. F. THOMAS & M. B. THORP
forests outward from their refugia or heartland areas. But the lake level data do not suggest that the severity of climate was everywhere as great as during the LGM. Hillslope erosion would have continued and colluviation may have been widespread at this time, but fluvial activity was diminished. The Holocene pluvial phase post 10 ka aP led to much higher lake levels than before and to renewed fluvial activity on a scale comparable with the late Pleistocene in some areas, but apparently did not lead to major sedimentation events in lower reaches and off-shore deltas. This has been interpreted as an indication of the stabilisation of the landscape as rainforest and other woodland vegetation was finally re-established between 9.5 and 8.5kaBP, according to location and site conditions. But it must also reflect the build-up of sediment stores in early Holocene floodplains, where both lateral and vertical accretion deposits remain stored in large volumes. This also seems to have been a period of widespread mass movement and colluviation. Palaeolandslides are common features throughout much of the humid tropics, but it is not known when these were first initiated and landslides are seldom dated. Nevertheless, an impression is gained of landscapes that have experienced sometime in the Holocene a phase of quite unparalleled morphogenetic activity which was triggered by a concentration of high magnitude rainfall events.
///,~
X F 2 ~,
YOUNGER FLOODPLAIN FORMATION
........................
I
I
I
14
13
12
...............................................
I~ 11
, ....................................................................................................................
~'
I
I
I
10.5
10
9
8
.....................
I
I
I
I
~
7
6
5
4
3.5
'
I
I
I
3
2
1
Kyr
Fig. 1. Late Quaternary stream activity in the humid tropics of West Africa Curve peaks denote timing of major sedimentary units; troughs/arrows indicate likely periods of channel cutting. Diagonal shading indicates periods of dry climate. Reprinted from Quaternary Science Reviews, 14, Thomas & Thorp, pp. 193-207, 1995, with kind permission from Elsevier Science Ltd, The Boulevard, Langford Lane, Kidlington OX5 1GB, UK.
149
HYDROLOGICAL CHANGE IN HUMID/SUB-HUMID TROPICS
/
\
/
/
/
~~~1'/
40
20
'~;30
20
" 10
\
